fix(types): four mypy errors blocking CI

refactor(harness): migrate to per-dimension Judge layer (drop faithfulness/correctness)

fix: comparison framing, mock-specific failure analysis, stale test counts

fix: grounded refusal checks no-sources, reference_answer for judge, mock disclaimer

fix: retrieval metrics use ranked sources, LLM judge wired, report complete

feat: Day 7 — evaluation harness, metrics, report, expanded golden dataset