Introduce test-reviewer agent into code review process

[NTCoding]

References

PR #259 reflection — CodeRabbit caught missing rename test that local reviewers missed

Summary

Split test review responsibilities from the main code-review agent into a dedicated test-reviewer agent. The current code-review agent handles both design/convention enforcement and test quality assessment, making it less focused on each. A dedicated test-reviewer would specialize in test completeness, edge case coverage, and format-specific test requirements like RFC-017.

Full Details

Create a new .claude/agents/test-reviewer.md agent that focuses exclusively on test quality. Include rules for: (1) external format parser test completeness (all documented variants), (2) edge case coverage per RFC-014, (3) success-path assertions per RFC-013, (4) mock cleanup per RFC-016. Wire it into the code review step in run-code-review.ts alongside existing reviewers.

Acceptance Criteria

• New .claude/agents/test-reviewer.md agent created with focused test quality rules
• Agent includes RFC-017 structured format parsing test requirement
• Agent wired into run-code-review.ts reviewer list
• Existing code-review agent updated to remove overlapping test responsibilities
• All tests pass with new reviewer in pipeline