chore: gitignore tier-4 corpus research + .terrain scratch

[pmclSF]

Summary

• Ignores raw corpus graph data in `tier-4/` (walks, boundaries, labels)
• Allowlists the small artifacts that ARE legitimate documentation: markdown reports, hand-labeled TSVs, repo URL lists
• Ignores `.terrain/snapshots/` and `.terrain/cache/` (per-run scratch)
• Adds `/terrain-precision` to match the existing `/terrain-corpus` rule

Why

The tier-4/ directory holds the harvested graph used for detector calibration. Without an ignore rule, accidental `git add tier-4/` sweeps in 300+ MB JSONL files (already triggered a push rejection during the README PR work). These are research scratch — they belong on the local machine that runs the harvest, never on GitHub.

No tier-4 data files are currently on origin/main. This is forward-looking only.

Test plan

• After merge, `git status` in a fresh checkout shows no tier-4/*.jsonl as untracked
• Existing tracked tier-4 markdown reports stay tracked (allowlist works)
• `.terrain/snapshots/` ignored

🤖 Generated with Claude Code

[github-actions]

[INFO] Terrain — Informational only

Insufficient data to assess change risk confidently.

Metric	Value
Changed files	1 (0 source · 0 test)

---

✓ All clear. No new findings introduced; no protection gaps identified in changed code.

Run terrain compare over time to track posture; this clean state is the bar to hold.

---

Limitations

• No coverage artifacts provided; protection gaps reflect missing data, not measured absence. Provide --coverage to improve accuracy.
• Mixed test cultures reduce cross-framework optimization confidence. Consider standardizing on fewer frameworks.

_{Generated by Terrain · terrain pr --json for machine-readable output}

Targeted Test Results

No tests selected — change affects only non-code files.

[github-actions]

Terrain AI Risk Review

Metric	Value
AI surfaces	13
Eval scenarios	17
Impacted scenarios	0
Uncovered surfaces	13

Decision: PASS — AI surfaces are covered.