[cschanhniem]

Implements Q4 from NEXT_100_PR_MAP.

What

• src/openamp_foundry/evidence/hit_confirmation_report.py: HCR- schema; HitConfirmationReport dataclass (16 fields); validate_hit_confirmation_report(); build_hit_confirmation_report() (enforces dry_lab_only=False); format_hit_confirmation_report()
• VALID_CONFIRMATION_VERDICTS (4): confirmed_hit, partial_hit, not_confirmed, inconclusive
• VALID_DIVERGENCE_TYPES (7): mic_above_predicted_range, mic_below_predicted_range, activity_spectrum_mismatch, selectivity_mismatch, hemolysis_above_predicted, no_divergence, multiple_divergences
• VALID_PREDICTION_QUALITY_GRADES (5): A, B, C, D, N/A
• tests/evidence/test_hit_confirmation_report.py: 63 tests in 5 classes
• BASELINE: 8774 → 8837

Disconfirming pass

• dry_lab_only=True blocked (tested) — real experimental outcomes required
• TOY- candidate IDs blocked (tested) — no toy data in real HCR- records
• empty whr_ids blocked (tested) — at least one WHR- record required
• confirmed_hit requires n_confirmed >= 1 (tested)
• not_confirmed requires n_inactive >= 1 (tested)
• no_divergence cannot coexist with other divergence types (tested)
• counts exceeding n_experiments blocked (tested)