[codex] unskip passing zlib inventory parity

[andrewtdiz]

Linked issues

• Closes part of compat: wire the 42-module parity matrix into CI with per-module trend #812

Behavior cluster summary

This removes the stale test_parity_zlib entry from test-parity/known_failures.json. The generated zlib inventory probe and the curated node-suite/zlib suite now both pass on current origin/main with Node 26, so the skip-list should no longer classify zlib inventory as an expected module-inventory failure.

Why this is batched

This is intentionally a narrow cleanup because nearby inventory rows are not equally safe to remove: DNS oracle runs exit nonzero in this environment, HTTP/SQLite inventory probes can hang or compile slowly, and the open parity-matrix trend PR #4286 explicitly leaves known-failure row cleanup out of scope. Keeping this to the proven zlib row avoids mixing stale cleanup with unresolved inventory work.

Tests and checks

• npm exec --yes --package=node@26 -- bash -lc './run_parity_tests.sh --filter test_parity_zlib' -> 1 pass / 0 fail / 0 compile fail, report test-parity/reports/parity_report_20260603_203746.json
• npm exec --yes --package=node@26 -- bash -lc './run_parity_tests.sh --suite node-suite --module zlib' -> 57 pass / 0 fail / 0 compile fail, report test-parity/reports/parity_report_20260603_203804.json
• jq empty test-parity/known_failures.json
• cargo fmt --all -- --check
• git diff --check
• ./scripts/check_file_size.sh

Known limitations

Only test_parity_zlib is removed here. Other module-inventory known-failure rows need their own clean, deterministic pass evidence before removal.

Non-goals

• Runtime zlib behavior changes.
• Parity-matrix CI/trend-gate wiring from [codex] add parity matrix trend gate #4286.
• HTTP, DNS, SQLite, stream, TLS, or CI-env skip-list cleanup.