Remove pre-strict Cypher binder guards - #1432

Closes #1420.

Summary

• Retires the pre-strict Cypher binder compatibility branch: FrontendBinder.bind(strict_name_resolution=False) remains source-compatible but no longer restores loose binder semantics.
• Makes compile prepass / graph-constructor binding explicitly strict and removes binder dependency on the rollout resolver as a production strict/loose gate.
• Updates rollout docs/tests and strict-mode docs to reflect [DEL] D2 remove pre-strict binder compatibility guards after #1357 #1420 strict-canonical behavior, including diagnostics that point users at query/schema fixes instead of the retired loose-mode escape hatch.
• Adds adversarial tests proving the legacy false flag no longer admits unresolved collect(...), single-alias list literal, or missing-schema inputs, while preserving source-order WITH -> UNWIND -> MATCH binding.
• Keeps WITH ... WHERE validation aligned with Cypher semantics by checking both projected aliases and source-scope aliases; this covers the CI/TCK WITH a.name2 AS name WHERE name = ... OR a.name2 = ... shape.

Validation

• python3 -m pytest -q graphistry/tests/compute/gfql/cypher/test_binder.py::test_binder_strict_schema_rejects_missing_match_label_with_context graphistry/tests/compute/gfql/cypher/test_binder.py::test_binder_strict_schema_rejects_missing_property_in_return_with_context graphistry/tests/compute/gfql/cypher/test_binder.py::test_binder_strict_schema_rejects_missing_relationship_property_in_match_pattern (3 passed)
• python3 -m pytest -q graphistry/tests/compute/gfql/cypher/test_binder.py graphistry/tests/compute/gfql/cypher/test_binder_strict_compile_baseline.py graphistry/tests/compute/gfql/cypher/test_validator_runtime_strict_parity.py graphistry/tests/compute/gfql/cypher/test_m1_differential_scaffold.py graphistry/tests/compute/gfql/test_rollout_binder_integration.py graphistry/tests/compute/gfql/test_rollout.py (151 passed, 29 warnings)
• python3 -m pytest -q graphistry/tests/compute/gfql/cypher/test_lowering.py -k "(strict or binder or unresolved or collect or unwind) and not cudf" (41 passed, 958 deselected, 6 warnings)
• python3 -m pytest -q graphistry/tests/compute/gfql/cypher/test_binder.py::test_binder_with_where_validates_against_projected_alias_scope graphistry/tests/compute/gfql/cypher/test_binder.py::test_binder_with_where_validates_against_source_alias_scope graphistry/tests/compute/gfql/cypher/test_binder.py::test_binder_with_where_rejects_unknown_source_or_projected_alias graphistry/tests/compute/gfql/cypher/test_lowering.py::test_string_cypher_supports_with_where_using_projected_source_properties (5 passed, 4 warnings)
• python3 -m pytest -q graphistry/tests/compute/gfql/test_logical_planner.py graphistry/tests/compute/gfql/cypher/test_binder.py graphistry/tests/compute/gfql/cypher/test_binder_strict_compile_baseline.py graphistry/tests/compute/gfql/cypher/test_validator_runtime_strict_parity.py graphistry/tests/compute/gfql/test_rollout_binder_integration.py (129 passed, 28 warnings)
• python3 -m pytest -q graphistry/tests/compute/gfql/cypher/test_lowering.py graphistry/tests/compute/gfql/cypher/test_parser.py graphistry/tests/compute/gfql/test_row_pipeline_ops.py graphistry/tests/compute/gfql/test_schema_changers.py graphistry/tests/compute/gfql/test_let_schema_changers.py graphistry/tests/compute/gfql/test_rollout.py graphistry/tests/compute/gfql/test_rollout_binder_integration.py graphistry/tests/compute/test_hop.py graphistry/tests/compute/test_chain.py graphistry/tests/compute/test_chain_let.py graphistry/tests/compute/test_chain_concat.py graphistry/tests/compute/test_dataframe_primitives.py tests/gfql/ref/ -k "not cudf" (1897 passed, 1 unrelated local UMAP/cuDF no-GPU failure)
• python3 -m py_compile graphistry/compute/gfql/frontends/cypher/binder.py graphistry/compute/gfql/cypher/lowering.py graphistry/compute/gfql/rollout.py graphistry/tests/compute/gfql/cypher/test_binder.py graphistry/tests/compute/gfql/cypher/test_m1_differential_scaffold.py graphistry/tests/compute/gfql/test_rollout_binder_integration.py graphistry/tests/compute/gfql/cypher/test_binder_cross_kind_rebind.py
• python3 -m py_compile graphistry/compute/gfql/frontends/cypher/binder.py graphistry/tests/compute/gfql/cypher/test_binder.py
• ./bin/ruff.sh graphistry/compute/gfql/frontends/cypher/binder.py graphistry/compute/gfql/cypher/lowering.py graphistry/compute/gfql/rollout.py graphistry/tests/compute/gfql/cypher/test_binder.py graphistry/tests/compute/gfql/cypher/test_m1_differential_scaffold.py graphistry/tests/compute/gfql/test_rollout_binder_integration.py graphistry/tests/compute/gfql/cypher/test_binder_cross_kind_rebind.py
• ./bin/ruff.sh graphistry/compute/gfql/frontends/cypher/binder.py graphistry/tests/compute/gfql/cypher/test_binder.py
• DGX ssh dgx-spark, RAPIDS 25.02 harness, focused binder + rollout + selected cuDF lowering nodes: 92 passed, 17 warnings
• DGX ssh dgx-spark, RAPIDS 26.02 harness, same focused selection: 92 passed, 16 warnings
• DGX ssh dgx-spark, direct 26.02 Docker rerun of the four cuDF nodes that had OOM'd under shared GPU pressure: 4 passed, 4 warnings
• DGX ssh dgx-spark, commit 23a7a559, RAPIDS 25.02 harness, focused binder + rollout + logical-planner repro + WITH WHERE source/projection + selected cuDF lowering nodes: 98 passed, 17 warnings
• DGX ssh dgx-spark, commit 23a7a559, RAPIDS 26.02 harness, same focused selection: 98 passed, 16 warnings

Note: a direct 26.02 no-harness run over the 92-test focused selection reached 88 passed before four cuDF tests hit std::bad_alloc on tiny allocations. nvidia-smi showed a concurrent VLLM::EngineCore using 81870MiB; the same four nodes passed when rerun directly after the failed container exited.