kubevirt: helm lint fails on the chart because _cluster is missing from its values #4030

What happens

helm lint packages/system/kubevirt fails on main:

[ERROR] templates/: template: cozy-kubevirt/templates/vm-exportproxy-tlsroute.yaml:1:13: executing "cozy-kubevirt/templates/vm-exportproxy-tlsroute.yaml" at <index .Values._cluster "root-host">: error calling index: index of untyped nil

templates/vm-exportproxy-tlsroute.yaml and templates/vm-exportproxy-ingress.yaml both start with index .Values._cluster "root-host", and the chart's own values.yaml has no _cluster key. The platform always injects _cluster at install time, so the deployed chart is fine; a standalone helm lint or helm template without --set _cluster.root-host=… is not. Nothing in CI runs lint on this chart, so nothing is red; it only bites whoever renders the chart by hand or adds a lint step later.

Fix shape

Either declare _cluster: {} in values.yaml and read the keys through dig/default so the two templates render nothing when the platform context is absent, or guard both templates with {{- with .Values._cluster }}. Same shape as the nil-safe walk the platform bundle already uses for its own optional blocks.

Seen while running the chart's unit tests for #4015; present at that PR's merge base and unrelated to it.