[system] Add Gateway API support via Cilium - #2208

Register a dedicated cozystack.gateway-api-crds PackageSource and wire
it into the system bundle before networking/cert-manager load. Without
this the host cluster has no gateway.networking.k8s.io CRDs, Cilium's
Gateway API controller cannot bootstrap (no GatewayClass registered),
and cert-manager's gateway-shim controller silently disables Gateway
API certificate automation.

- New PackageSource cozystack.gateway-api-crds → installs
  packages/system/gateway-api-crds into namespace cozy-gateway-api.
- cozystack.networking (variants cilium, cilium-kilo, cilium-generic,
  kubeovn-cilium, kubeovn-cilium-generic) dependsOn
  cozystack.gateway-api-crds so Cilium only starts once the CRDs are
  present.
- cozystack.cert-manager dependsOn cozystack.gateway-api-crds, closing
  the startup race that the #2208 discussion flagged — cert-manager's
  gateway-shim discovers CRDs at process start, so racing with CRD
  install left Gateway API solvers permanently disabled until the pod
  was restarted.
- system bundle loads cozystack.gateway-api-crds unconditionally for
  every variant (isp-full, isp-hosted, isp-full-generic).

Assisted-By: Claude <[email protected]>
Signed-off-by: Aleksei Sviridkin <[email protected]>

Register a dedicated cozystack.gateway-api-crds PackageSource and wire
it into the system bundle before networking/cert-manager load. Without
this the host cluster has no gateway.networking.k8s.io CRDs, Cilium's
Gateway API controller cannot bootstrap (no GatewayClass registered),
and cert-manager's gateway-shim controller silently disables Gateway
API certificate automation.

- New PackageSource cozystack.gateway-api-crds → installs
  packages/system/gateway-api-crds into namespace cozy-gateway-api.
- cozystack.networking (variants cilium, cilium-kilo, cilium-generic,
  kubeovn-cilium, kubeovn-cilium-generic) dependsOn
  cozystack.gateway-api-crds so Cilium only starts once the CRDs are
  present.
- cozystack.cert-manager dependsOn cozystack.gateway-api-crds, closing
  the startup race that the #2208 discussion flagged — cert-manager's
  gateway-shim discovers CRDs at process start, so racing with CRD
  install left Gateway API solvers permanently disabled until the pod
  was restarted.
- system bundle loads cozystack.gateway-api-crds unconditionally for
  every variant (isp-full, isp-hosted, isp-full-generic).

Assisted-By: Claude <[email protected]>
Signed-off-by: Aleksei Sviridkin <[email protected]>

Register a dedicated cozystack.gateway-api-crds PackageSource and wire
it into the system bundle before networking/cert-manager load. Without
this the host cluster has no gateway.networking.k8s.io CRDs, Cilium's
Gateway API controller cannot bootstrap (no GatewayClass registered),
and cert-manager's gateway-shim controller silently disables Gateway
API certificate automation.

- New PackageSource cozystack.gateway-api-crds → installs
  packages/system/gateway-api-crds into namespace cozy-gateway-api.
- cozystack.networking (variants cilium, cilium-kilo, cilium-generic,
  kubeovn-cilium, kubeovn-cilium-generic) dependsOn
  cozystack.gateway-api-crds so Cilium only starts once the CRDs are
  present.
- cozystack.cert-manager dependsOn cozystack.gateway-api-crds, closing
  the startup race that the #2208 discussion flagged — cert-manager's
  gateway-shim discovers CRDs at process start, so racing with CRD
  install left Gateway API solvers permanently disabled until the pod
  was restarted.
- system bundle loads cozystack.gateway-api-crds unconditionally for
  every variant (isp-full, isp-hosted, isp-full-generic).

Assisted-By: Claude <[email protected]>
Signed-off-by: Aleksei Sviridkin <[email protected]>

Register a dedicated cozystack.gateway-api-crds PackageSource and wire
it into the system bundle before networking/cert-manager load. Without
this the host cluster has no gateway.networking.k8s.io CRDs, Cilium's
Gateway API controller cannot bootstrap (no GatewayClass registered),
and cert-manager's gateway-shim controller silently disables Gateway
API certificate automation.

- New PackageSource cozystack.gateway-api-crds → installs
  packages/system/gateway-api-crds into namespace cozy-gateway-api.
- cozystack.networking (variants cilium, cilium-kilo, cilium-generic,
  kubeovn-cilium, kubeovn-cilium-generic) dependsOn
  cozystack.gateway-api-crds so Cilium only starts once the CRDs are
  present.
- cozystack.cert-manager dependsOn cozystack.gateway-api-crds, closing
  the startup race that the #2208 discussion flagged — cert-manager's
  gateway-shim discovers CRDs at process start, so racing with CRD
  install left Gateway API solvers permanently disabled until the pod
  was restarted.
- system bundle loads cozystack.gateway-api-crds unconditionally for
  every variant (isp-full, isp-hosted, isp-full-generic).

Assisted-By: Claude <[email protected]>
Signed-off-by: Aleksei Sviridkin <[email protected]>

Register a dedicated cozystack.gateway-api-crds PackageSource and wire
it into the system bundle before networking/cert-manager load. Without
this the host cluster has no gateway.networking.k8s.io CRDs, Cilium's
Gateway API controller cannot bootstrap (no GatewayClass registered),
and cert-manager's gateway-shim controller silently disables Gateway
API certificate automation.

- New PackageSource cozystack.gateway-api-crds → installs
  packages/system/gateway-api-crds into namespace cozy-gateway-api.
- cozystack.networking (variants cilium, cilium-kilo, cilium-generic,
  kubeovn-cilium, kubeovn-cilium-generic) dependsOn
  cozystack.gateway-api-crds so Cilium only starts once the CRDs are
  present.
- cozystack.cert-manager dependsOn cozystack.gateway-api-crds, closing
  the startup race that the #2208 discussion flagged — cert-manager's
  gateway-shim discovers CRDs at process start, so racing with CRD
  install left Gateway API solvers permanently disabled until the pod
  was restarted.
- system bundle loads cozystack.gateway-api-crds unconditionally for
  every variant (isp-full, isp-hosted, isp-full-generic).

Assisted-By: Claude <[email protected]>
Signed-off-by: Aleksei Sviridkin <[email protected]>

Register a dedicated cozystack.gateway-api-crds PackageSource and wire
it into the system bundle before networking/cert-manager load. Without
this the host cluster has no gateway.networking.k8s.io CRDs, Cilium's
Gateway API controller cannot bootstrap (no GatewayClass registered),
and cert-manager's gateway-shim controller silently disables Gateway
API certificate automation.

- New PackageSource cozystack.gateway-api-crds → installs
  packages/system/gateway-api-crds into namespace cozy-gateway-api.
- cozystack.networking (variants cilium, cilium-kilo, cilium-generic,
  kubeovn-cilium, kubeovn-cilium-generic) dependsOn
  cozystack.gateway-api-crds so Cilium only starts once the CRDs are
  present.
- cozystack.cert-manager dependsOn cozystack.gateway-api-crds, closing
  the startup race that the #2208 discussion flagged — cert-manager's
  gateway-shim discovers CRDs at process start, so racing with CRD
  install left Gateway API solvers permanently disabled until the pod
  was restarted.
- system bundle loads cozystack.gateway-api-crds unconditionally for
  every variant (isp-full, isp-hosted, isp-full-generic).

Assisted-By: Claude <[email protected]>
Signed-off-by: Aleksei Sviridkin <[email protected]>

Register a dedicated cozystack.gateway-api-crds PackageSource and wire
it into the system bundle before networking/cert-manager load. Without
this the host cluster has no gateway.networking.k8s.io CRDs, Cilium's
Gateway API controller cannot bootstrap (no GatewayClass registered),
and cert-manager's gateway-shim controller silently disables Gateway
API certificate automation.

- New PackageSource cozystack.gateway-api-crds → installs
  packages/system/gateway-api-crds into namespace cozy-gateway-api.
- cozystack.networking (variants cilium, cilium-kilo, cilium-generic,
  kubeovn-cilium, kubeovn-cilium-generic) dependsOn
  cozystack.gateway-api-crds so Cilium only starts once the CRDs are
  present.
- cozystack.cert-manager dependsOn cozystack.gateway-api-crds, closing
  the startup race that the #2208 discussion flagged — cert-manager's
  gateway-shim discovers CRDs at process start, so racing with CRD
  install left Gateway API solvers permanently disabled until the pod
  was restarted.
- system bundle loads cozystack.gateway-api-crds unconditionally for
  every variant (isp-full, isp-hosted, isp-full-generic).

Assisted-By: Claude <[email protected]>
Signed-off-by: Aleksei Sviridkin <[email protected]>

…2470)

## What this PR does

Replaces ingress-nginx as the default ingress layer for cozystack-native
services with **Cilium Gateway API**, materialised per-tenant via a new
`TenantGateway` CRD reconciled by `cozystack-controller`. Supersedes
#2213 / #2208. Stacked on top of #2468.

## Architecture

The chart renders one `gateway.cozystack.io/v1alpha1 TenantGateway` CR
per tenant that owns its own Gateway. The cozystack-controller
reconciles the actual `Gateway`, per-tenant `Issuer`, `Certificate`
(wildcard or per-listener), and HTTP→HTTPS redirect `HTTPRoute` from
there. Helm does not render `Gateway` or `Certificate` directly — that
prevents the Helm-vs-controller race on `Gateway.spec.listeners` that
route-driven listener additions would otherwise cause.

```
┌── TenantGateway CR (rendered by extra/gateway chart for owning tenants)
│
└──▶ cozystack-controller reconciles:
        ├─ Gateway (with dynamic listeners, label-based allowedRoutes)
        ├─ Issuer (per-tenant ACME account)
        ├─ Certificate(s) — wildcard + per-child SANs in DNS-01, per-listener in HTTP-01
        ├─ HTTPRoute (controller-owned http→https redirect)
        ├─ Namespace label patching (cozy-* system namespaces from spec.AttachedNamespaces)
        └─ Watches HTTPRoute / TLSRoute attachments to materialise listeners
```

## Inheritance: tenants attach by default, opt into their own Gateway

A tenant gets a dedicated Gateway (own LB Service, own LB IP, own
Certificate) only when it explicitly asks via
`tenant.spec.gateway=true`. Every other tenant in the tree publishes
through the Gateway of the nearest ancestor that owns one — same shape
as the existing `_namespace.ingress` inheritance.

The `apps/tenant` chart writes a `namespace.cozystack.io/gateway` label
on each tenant namespace, carrying either the tenant's own name (when
owning a Gateway) or the inherited ancestor name (when inheriting). The
Gateway's listener `allowedRoutes.namespaces.selector` is keyed on that
label, so the same selector admits routes from every namespace pointing
at the owner — descendants and cozy-* system namespaces alike.
cozystack-controller patches the same label onto every namespace in
`tgw.Spec.AttachedNamespaces` and garbage-collects labels it wrote when
an entry is removed from the attach list (annotation-gated GC so labels
written by the apps/tenant chart are never stripped).

In DNS-01 mode, the controller extends the parent's wildcard Certificate
with `<child-apex>` + `*.<child-apex>` SANs per inheriting child, and
adds a `*.<child-apex>` HTTPS listener per child apex referencing the
same cert. Without this expansion the parent's single-level wildcard
cannot match a child route's hostname (`harbor.alice.example.org` is two
labels past the parent's `*.example.org`).

Opt-out into own Gateway is the right choice when: a tenant needs its
own LB IP (DNS pinned, firewall rule), the apex is not derived from the
parent (custom `host` value the ancestor's cert can't cover), or the
tenant wants its own ACME account / cert authority. Otherwise leave
`gateway` unset and inherit.

## End-to-end flow

1. Platform: `gateway.enabled=true` toggles cert-manager solver,
switches per-system-app templates from Ingress to HTTPRoute / TLSRoute.
2. Per tenant: explicit `tenant.spec.gateway=true` renders a
`TenantGateway` CR and triggers the controller to materialise the owner
Gateway. Tenants with the field unset inherit through
`_namespace.gateway` propagation in the tenant chart.
3. Adding a published app under any tenant (owner or inheriting) is just
deploying its HTTPRoute. The controller picks up the hostname, adds a
per-listener Certificate (HTTP-01) or extends the wildcard cert's SANs
(DNS-01), and updates the owner Gateway's listeners.

Defaults stay legacy: `gateway.enabled=false`,
`tenant.spec.gateway=false`. Existing clusters upgrade unchanged.

## External IP allocation — mechanism-agnostic

The per-tenant Gateway's auto-created `LoadBalancer` Service draws its
IP from whatever LB allocator the cluster admin has configured at the
platform layer — same shape as ingress-nginx today. The tenant API stays
mechanism-agnostic — no `gatewayIP` field, no allocator-specific
manifest in the tenant chart. Cozystack itself ships MetalLB installed
but does not render any `IPAddressPool` / `L2Advertisement` /
`BGPAdvertisement` from this chart; admins set up the allocator that
suits their environment (MetalLB pool with L2 / BGP, Cilium LB-IPAM with
announcer, robotlb against a cloud provider, or
`Service.spec.externalIPs` pinning).

If a tenant needs a specific address (DNS already pinned, firewall rule,
etc.), the operator pre-allocates it on the admin side: either
pre-create the Service with `loadBalancerIP` set, or hand the tenant a
reference to a named admin-managed pool. Per-Service IP uniqueness is
the allocator's responsibility — same as for any other LoadBalancer
Service.

## Cert mode: HTTP-01 (default) vs DNS-01 (opt-in)

`publishing.certificates.solver` selects how the controller sources TLS
certs.

**HTTP-01 (default)** — per-app HTTPS listener + per-app Certificate.
New apps require zero platform-side config: deploy the HTTPRoute,
controller does the rest. Inheriting children's hostnames pick up
per-listener certs the same way the owner's own apps do.

**DNS-01 (opt-in)** — single wildcard Certificate (`<apex>` +
`*.<apex>`) covering every published app under the apex, plus per-child
SANs (`<child-apex>` + `*.<child-apex>`) for every inheriting tenant.
Pick this for clusters where many apps share the apex and Let's Encrypt
rate limits matter. The DNS provider account must be able to write TXT
records under every apex level the parent serves — for deeply-nested
inherited children that requires zone delegation or a wide-scope
provider credential.

| `publishing.certificates.dns01.provider` | Required
`publishing.certificates.dns01.<provider>` keys |
| --- | --- |
| `cloudflare` (default) | `cloudflare.secretName`,
`cloudflare.secretKey` |
| `route53` | `route53.region`, `route53.secretName` (and `accessKeyID`
if not on IRSA) |
| `digitalocean` | `digitalocean.secretName` |
| `rfc2136` | `rfc2136.nameserver`, `rfc2136.tsigKeyName`,
`rfc2136.secretName` |

The platform chart wires every provider's keys into `_cluster.dns01-*`;
the per-tenant gateway chart and the cluster-wide `cluster-issuers.yaml`
both read them.

## Strengthens admission coverage on apps.cozystack.io/*

The `cozystack-tenant-host-policy` probe in `gateway.bats` initially
looked like a flake — the apply that was supposed to be VAP-rejected
just succeeded. Investigation traced it to the custom REST handler at
`pkg/registry/apps/application/rest.go`. genericapiserver hands every
storage method a callback into the validating admission chain:
`createValidation` for Create, `updateValidation` for Update,
`deleteValidation` for Delete. `Update` was correctly invoking its
callback since 23e399b, but `Create` and `Delete` accepted the
parameter and never called it — admission was silently bypassed for both
verbs on every `apps.cozystack.io` resource (Tenant, MariaDB, Postgres,
Kubernetes, …).

This PR wires the missing `createValidation` and `deleteValidation`
calls into Create and Delete. Every ValidatingAdmissionPolicy and
ValidatingWebhook against `apps.cozystack.io/*` now fires on all three
verbs as the API contract requires. Per KEP-3488 aggregated APIServers
are responsible for enforcing admission on their own kinds;
cozystack-api is configured to do so via
`genericoptions.NewRecommendedOptions`, the only thing missing was the
explicit `*Validation` callbacks in the custom REST.

## Security — five guards, grouped by what they defend

Tenants in Cozystack interact with the platform exclusively through
`apps.cozystack.io/*` resources (Tenant, Bucket, Kubernetes, …) served
by `cozystack-api`. Tenant RBAC (`cozy:tenant:*` aggregated to a
RoleBinding in the tenant's own namespace) does not grant write access
to `gateway.networking.k8s.io/*`, core `Namespaces`, or
`cozystack.io/Package`. The protections below split into three groups by
who they defend against:

**Tenant-user-input gates.** Layer 4 (`cozystack-tenant-host-policy`).
`Tenant.spec.host` is the user-supplied field that surfaces as a
security boundary at the hostname layer; gated on every Create / Update
via `cozystack-api`'s admission chain.

**Defense-in-depth.** Layers 1, 2, 5, 7. These do not protect against
tenant-user input (tenants don't hold the relevant RBAC). They guard
against bugs in cozystack-controller / Flux, supply-chain compromise of
an app chart that emits Gateway API resources, and confused-deputy
mistakes by a cluster admin. Fail-closed via `failurePolicy: Fail` +
`validationActions: [Deny]`.

The five layers themselves:

1. **Listener `allowedRoutes` label selector** —
`namespace.cozystack.io/gateway = <owner-tenant-name>`
(controller-written for cozy-* via `AttachedNamespaces`, Helm-written
for tenant namespaces via the tenant chart, both feeding into the same
Gateway selector).
2. **`cozystack-gateway-hostname-policy`** — VAP on `Gateway`
CREATE/UPDATE. Restricts listener hostnames to the namespace's
`namespace.cozystack.io/host` apex.
4. **`cozystack-tenant-host-policy`** — VAP on `Tenant` CREATE/UPDATE.
5. **`cozystack-namespace-host-label-policy`** — VAP on core `Namespace`
CREATE/UPDATE. Immutability gate on the host label.
7. **`cozystack-route-hostname-policy`** — VAP on `HTTPRoute` and
`TLSRoute` CREATE/UPDATE. Closes the hostname-hijack vector across
namespaces under inheritance — a child cannot claim a hostname outside
its own apex.

The previous Layer 3 (`cozystack-gateway-attached-namespaces-policy`)
and Layer 6 (render-time `fail` in cozystack-basics) banned `tenant-*`
entries in `publishing.gateway.attachedNamespaces` on the assumption
that the attach list was the hijack vector. Under inheritance the attach
surface is the label selector, not the static list; the hijack vector is
closed by Layers 4/5/7 independently. Both Layer 3 and Layer 6 are
dropped — kept in numbering above for continuity with the earlier review
threads.

## Foreign-takeover guards

Five reconcile paths refuse to silently rewrite a pre-existing object
that shares the controller-derived name but carries no `OwnerReference`
back to the TenantGateway: `Gateway`, redirect `HTTPRoute`, per-tenant
`Issuer`, wildcard `Certificate`, per-listener `Certificate`. An
operator who hand-pinned a Certificate or Issuer at the controller's
derived name (private CA, manual cert pinning) gets an explicit
`Ready=False/ReconcileError` condition instead of having their config
silently destroyed and the resource re-issued from a different ACME
account.

The namespace-label patching path applies the same discipline:
cozystack-controller only writes / strips
`namespace.cozystack.io/gateway` on namespaces it annotates with
`cozystack.io/gateway-attached-by`. Labels written by the apps/tenant
chart (no annotation) are never touched, so inheritance for tenant
namespaces survives every reconcile.

## Tests

- **Go controller tests** — 50+ cases in
`internal/controller/tenantgateway/`: idempotency, OwnerReference
cascade, multi-parentRef status, HTTP-01↔DNS-01 mode-transition cert
lifecycle, foreign-takeover refusals (Gateway / HTTPRoute / Issuer /
Certificate / per-listener Certificate), DNS-01 provider matrix
(cloudflare / route53 / digitalocean / rfc2136), HTTPS Kinds=[HTTPRoute]
restriction, fail-closed CEL on missing host label, status condition
wiring. Plus, for inheritance: label-based allowedRoutes selector shape,
namespace label patch + garbage-collect via owner annotation, Helm-owned
label safety, DNS-01 wildcard SAN expansion per inheriting child,
per-child wildcard listener materialisation.
- **Admission-chain tests** —
`pkg/registry/apps/application/rest_admission_test.go` pins
`createValidation`, `updateValidation`, `deleteValidation` paths via
sentinel callbacks.
- **helm-unittest** across `extra/gateway` (TenantGateway CR rendering,
both modes), `core/platform` (DNS-01 wiring), `apps/tenant`
(gateway-effective decision matrix including inheritance),
`system/cozystack-basics` (three VAPs + Layer 7 VAP + tenant-root
namespace label including the conditional
`namespace.cozystack.io/gateway: tenant-root` + cozystack-values-secret
`_namespace.gateway` propagation), `system/cert-manager-issuers` (full
DNS-01 provider matrix), `system/cozystack-api` and `system/dashboard`
(Ingress↔TLSRoute/HTTPRoute toggle), `system/monitoring` (Grafana /
Alerta Ingress↔HTTPRoute toggle).
- **e2e bats** — `hack/e2e-apps/gateway.bats` (17 cases) covering
GatewayClass Accepted, Gateway Programmed + LB Service, HTTPRoute
Accepted with status, VAP rejects cross-tenant hostname hijack, VAP
rejects `tenant.spec.host` writes from non-trusted SA
(regression-detector for the admission-chain fix), VAP rejects
`namespace.cozystack.io/host` label changes from non-trusted SA,
route-hostname VAP rejects HTTPRoute hostnames outside the apex, and
three new cases for inheritance — Package admission accepts `tenant-*`
in `attachedNamespaces` (regression guard for the dropped ban), child
tenant without explicit gateway inherits `_namespace.gateway` from a
Gateway-owning parent in lockstep across the namespace label and the
cozystack-values Secret, and the cross-namespace attach end-to-end where
a child tenant's HTTPRoute reaches Accepted on the parent's Gateway via
the inheritance label.
- **CI wiring** — `make test-controllers` target invoked from
`pull-requests.yaml` runs `go test ./internal/...`.

## Backward compatibility

The only always-on change is the flip of `cilium.envoy.enabled` +
`cilium.gatewayAPI.enabled` to `true` (extra `cilium-envoy` DaemonSet,
~100 MB RAM/node idle). Everything else is gated behind
`gateway.enabled=false` / `tenant.spec.gateway=false`. Existing clusters
see no behavioural change until an operator explicitly opts in.

The cozystack-api admission-chain fix is unconditional. Existing
operators who do not define any VAP / admission webhook on
`apps.cozystack.io` resources see no observable change. Operators who DO
have their own webhooks/VAPs against these kinds will start seeing them
fire on Create and Delete (as intended) — review your policies before
merging if you maintain custom admission for `apps.cozystack.io/*`.

Operators who set `publishing.gateway.attachedNamespaces` to include
`tenant-*` entries used to hit a render-time fail or an admission
rejection — now the apply succeeds and the listed tenant namespaces
simply pick up the gateway-attach label alongside the cozy-* system
namespaces. This matches the inheritance model and was kvaps' explicit
ask in the May 25 review.

## Release note

```release-note
- Adds Gateway API support via Cilium, materialised per-tenant via the new `gateway.cozystack.io/v1alpha1 TenantGateway` CRD reconciled by `cozystack-controller`. Opt-in via `publishing.gateway.enabled=true` at the platform level; per-tenant opt-in via `tenant.spec.gateway=true` for dedicated Gateway / LB IP, otherwise the tenant inherits the nearest ancestor's Gateway via a label-based `allowedRoutes` selector (same shape as the existing `_namespace.ingress` inheritance). Includes HTTP-01 (default) and DNS-01 (opt-in, with cloudflare / route53 / digitalocean / rfc2136 providers) cert solvers; DNS-01 extends the parent's wildcard Certificate with per-child apex SANs and adds a `*.<child-apex>` listener per inheriting tenant. Tenant API is mechanism-agnostic for IP allocation — the LoadBalancer Service draws its IP from the cluster's admin-configured allocator (MetalLB / Cilium LB-IPAM / robotlb / externalIPs).
- Strengthens admission coverage on `apps.cozystack.io/*` resources: `cozystack-api` now invokes `createValidation` and `deleteValidation` callbacks on Create and Delete (Update was already wired). Operators with custom ValidatingAdmissionPolicies or webhooks targeting `apps.cozystack.io/*` will see them fire on all three verbs as the API contract requires; review your policies before merging if you maintain such admission.
```