feat: Cross-encoder search reranking with MiniLM-L-6

Phase 6.5: Pipeline telemetry, search UX fixes, latency profiling

Phase 6.5: Documentation finalization

Phase 6.5 Day 5: Semantic Scholar author import (B4)

Phase 6.5 Day 4: Cluster snapshot versioning (B3)

Phase 6.5 Day 3: Propensity logging (B2)

Phase 6.5 Day 2: query_id linkage (B1)

Phase 6.5 Day 1: Real Qdrant cosine scores (A1) + verification timestamp (A2)

fix(onboarding): correct hybrid_search call + add CLAUDE.md Rule 3.10

Phase 6.3: Bug B fix + documentation finalization

Phase 6.1+6.2+6.3: Wire all 37 features into caller + health endpoint

Phase 6: LightGBM reranker integration (37-feature schema, 141-tree model)

Phase 6: Add arxiv_ids export script (IDs file on HF model repo)

Phase 5: Cold-Start Onboarding + UI Redesign

Phase 4 complete + Phase 4.5 instrumentation foundation

docs: Update tracker - Phase 3.5 fully complete, all integration done

Phase 3.5: Wire Turso DB for metadata (2.9x faster, includes citations)

docs: Add Phase 3.5 Turso ArXiv Metadata DB, update infra status

Fix SQLite permissions: writable dir + /tmp DB path for non-root user

Add python-multipart for FastAPI form endpoints

Pin transformers<5.0 for FlagEmbedding compat, add python-dotenv

Add HF Spaces README with Docker SDK metadata

Use python-dotenv for secrets management

Phase 3 complete: Hybrid Semantic Search pipeline