ponytail-audit: no pass to flag AI-shaped test brittleness (fixture-name branches in prod, snapshot/mock-only assertions, 1:1 method-mirror tests)

[Lockyer228]

Why

ponytail's test stance is "non-trivial logic leaves one runnable check," but ponytail-audit — scanning existing code — has no pass for AI-shaped test slop. Distinct from #602 (which constrains the agent while writing tests), this is about detecting brittleness in tests that already exist.

What

Add a test: tag and a Hunt entry flagging three failure modes:

• tests that mirror a method 1:1 (break on any refactor; they assert structure, not behavior)
• snapshot-only or mock-call-only assertions (they pin the AI's own output, not observable behavior)
• worst: fixture-name / id / timestamp branches leaking into production code to make a test pass

Plus a Boundaries clarification that test brittleness is dead weight to cut, not a correctness fix — keeping audit's "correctness out of scope" stance intact.

Tests

• node --test tests/openclaw-skills.test.js → 18/18
• node scripts/check-rule-copies.js → exit 0
• npm test → all pass except the pre-existing csv: correct pandas one-liner passes failure (fails identically on base; no pandas installed locally) — unrelated to this change.

Fixes #683