checks: Expand out-of-service detection to fabric ports

[monrog2]

Summary

• include fabric ports in the Out-of-Service Ports validation
• match usage masks 32, 34, 36, and 292
• use explicit equality filters so unrelated usage-mask values are not selected
• update regression fixtures, assertions, and validation documentation

Validation

• git diff --check origin/v4.2.0-dev...HEAD passed
• targeted pytest was not run locally because pytest is unavailable in the current environment

Fixes #292

[monrog2]

Integration test results

Tested commit 152b77f using the collect-validation-bundles integration runner via GitLab pipeline 172 / job 515.

• Python 2.7 unit test job: passed
• Python 3.8 unit test job: passed
• Validation bundles collected: 15 TGZ files
• PR-specific out_of_service_ports_check: passed on all 15 fabrics
• Local targeted regression suite: 12/12 passed
• Local full regression suite: 926/926 passed

The integration job exited nonzero because of unrelated baseline and lab issues. The primary script error was stale_dbgacEpgSummaryTask_check referencing timedelta without importing it on 12 fabrics; the same defect is present on the v4.2.0-dev base. Other errors were APIC authentication, API timeout/failure, or malformed API response issues. Three fabrics failed before producing bundles, and four unreachable fabrics were ignored by the runner.

No integration failure was attributable to this PR.

[monrog2]

No findings.

Reviewed the complete v4.2.0-dev...152b77f diff for APIC query correctness and scale, validation semantics and registration, Python 2.7/3.8 compatibility, documentation, and regression coverage. The explicit usage-mask filter matches issue #292 (32, 34, 36, and 292) without widening to unrelated masks, and remains a single server-side class query.

Validation evidence:

• GitHub Python 2.7 and 3.8 checks passed
• Targeted regression suite: 12/12 passed
• Full regression suite: 926/926 passed
• TGZ integration: out_of_service_ports_check passed on all 15 collected fabrics

The integration job failures were confirmed unrelated to this PR and present in the base or lab environment. This change is ready from a code-review perspective.

Submitted as a comment review because GitHub does not allow the PR author to approve their own PR.