feat: concurrent document indexing in index_folder (BRIC-8)

[Kaiohz]

Summary

Replace the sequential for loop in LightRAGAdapter.index_folder with asyncio.Semaphore + asyncio.gather to process documents concurrently, bounded by the existing MAX_CONCURRENT_FILES config.

Changes

• src/infrastructure/rag/lightrag_adapter.py: Replaced sequential loop with concurrent pattern using asyncio.Semaphore(max(1, MAX_CONCURRENT_FILES)) + asyncio.gather. Removed 103 lines of dead code.

• tests/unit/test_lightrag_adapter.py: Added 6 tests verifying: semaphore enforcement, zero-concurrency, excess capacity, single file, and mixed success/failure.

Acceptance Criteria (BRIC-8)

All 10 acceptance criteria verified.

Testing

• 127 unit tests pass (6 new + 121 existing)
• SonarQube: 0 new issues
• Trivy: 0 new vulnerabilities
• QA: Container healthy, folder indexing verified

Jira

BRIC-8

[Kaiohz]

🟢 Code Review — BRIC-8: Concurrent document indexing

✅ Points positifs

• Pattern asyncio.Semaphore + gather bien implémenté pour la concurrence
• 6 nouveaux tests (semaphore enforcement, zero-concurrency, excess capacity, single file, mixed success/failure)
• 103 lignes de dead code supprimées
• 127 tests passent, SonarQube clean

⚠️ Points d'attention

1. Fallback silencieux si MAX_CONCURRENT_FILES=0 → le max(1, ...) masque une config potentiellement intentionnelle à 0

Score de confiance : 8/10