feat(review): port reviewbot blockers[]/nits[] shape for extensive reviews

[JSONbored]

Why

The free Workers-AI pair was producing shallow, hedging reviews ("appears to…", "seems to…") even with the diff fully present. Confirmed via #1049: the stored file payloads contain real patches (has_patch=1, 5–6 KB each) and the AI re-ran on the re-run checkbox — so the diff was never the problem. The gap is the prompt structure.

reviewbot's gold-standard depth comes from forcing the model to enumerate findings into explicit blockers[] (concrete must-fix) + nits[] (non-blocking) arrays under a severity rubric — not an open-ended "be thorough".

What

Port reviewbot's recipe into src/services/ai-review.ts:

• REVIEW_SYSTEM_PROMPT → blockers[]/nits[]/suggestions arrays + severity discipline (a blocker points to a real diff defect; nits/hypotheticals never block; CI status is never a code finding), "do not rubber-stamp", no hedging.
• ModelReview / parseModelReview → parse blockers + nits (finding cap 6 → 12).
• composeAdvisoryNotes → render Blockers + Nits (nits + suggestions merged).
• consensusDefectOf → a consensus defect = a concrete blocker in both reviews (severity-disciplined; a lone blocker is a split, not a hard block) — replaces the numeric confidence floor.

Internal to ai-review.ts: the external advisoryNotes string + consensusDefect types are unchanged, so the gate and unified comment are unaffected.

Tests

Full suite green (3483 passed). Updated ai-review, ai-review-advisory, queue fixtures/assertions to the new shape; added a nits-don't-block consensus test.

[superagent-security]

Superagent didn't find any vulnerabilities or security issues in this PR.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.76%. Comparing base (01c315b) to head (9947d13).
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1089      +/-   ##
==========================================
- Coverage   94.76%   94.76%   -0.01%     
==========================================
  Files         153      153              
  Lines       18542    18540       -2     
  Branches     6711     6705       -6     
==========================================
- Hits        17572    17570       -2     
  Misses        418      418              
  Partials      552      552              

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.