feat: implement collaboration scoring enhancements — LLM sampling and human override #232
Description
Summary
Collaboration scoring has automated behavioral telemetry. Two enhancements from D3 are not implemented: periodic LLM sampling (1%) for calibration, and human override via API.
Design Spec Reference
- §8.3 Performance Tracking — D3
Scope
- Periodic LLM sampling (1% of interactions) for calibration against behavioral metrics
- Human override via API endpoint
- Integration with
CollaborationScoringStrategyprotocol