docs(skills): improve reference navigation and context hygiene

[sent-dm]

What changed

• Added complete linked TOCs to both canonical references over 100 lines.
• Added validator coverage for required TOCs, every internal anchor, skill-local bundled paths, one-H1 reference structure, and deprecated Suggested bundled... titles.
• Removed the MDR skill body's Unverified claims audit section.
• Moved necessary caveats and provenance into reference-level source notes.
• Added schemas/sent/documentation-sources.json with URL, kind, affected skills, and last_verified metadata for the freshness monitor.
• Added source-catalog contract tests and five new rejection gates.

Why

Long references were difficult to navigate, anchor drift was unchecked, and provenance caveats were loaded in the core workflow instead of living with the detailed claims they qualify.

Validation

• python3 scripts/generate_adapters.py --check
• python3 scripts/validate.py
• python3 scripts/test_fixtures.py
• python3 scripts/test_contracts.py
• python3 scripts/test_validation_gates.py
• Agent Skills validation for all canonical skills
• Strict Claude marketplace and plugin validation
• git diff --check

Stacked on the discovery metadata PR.