Ralph Loop 2026-03-16: Matched Research Sleeve Concentration

Bottleneck

research/latest_state.json showed candidate_review.counts_by_state.research_candidate=18, but only matched_research_candidate_count=3. The packet already split matched blocker tallies from unmatched inventory, but it did not expose matched-only sleeve concentration. That made it harder to tell whether the valid research queue was concentrated in one sleeve family or merely diluted by unmatched rows.

Change

Added top_matched_research_candidate_sleeves to the candidate-review packet and threaded the top matched research sleeve through:

• research/candidate_review.py
• research/continuous_loop.py
• research/status.py
• research/discord_status.py

This keeps unmatched inventory diagnostics intact while making the semantically valid sleeve mix visible in loop events and operator summaries.

Validation

Ran:

pytest tests/test_candidate_review.py tests/test_continuous_loop.py tests/test_research_status.py

Result: 39 passed in 0.19s