Problems with separate evaluations for real user data and llm simulated data

[yuanwuyuan9]

• Cause: Inconsistent JSON formats between real-user data and LLM-Simulation data.
• Solution: Implemented two separate processing pipelines:
  • Pipeline 1 (For Real Users):
    • coverage_metrics.py
    • cohesion_metrics.py
    • main_evaluator.py
  • Pipeline 2 (For LLM Simulation):
    • coverage_metrics_llm.py
    • cohesion_metrics_llm.py
    • main_evaluator_llm.py