Add ML Flow

[ccs-gs]

Summary

• Added optional MLflow tracking to evaluation/run_evaluation.py (modeled after the WIS evaluation flow) so evaluations can log:
  • run params (mapper, truth set path, prompt name/path, sample count)
  • run metrics (accuracy_percent, accuracy_fraction, correct_predictions, evaluation_duration_seconds)
  • artifacts (selected prompt file and output results CSV)
• Added new CLI flags for MLflow configuration:
  • --mlflow
  • --mlflow-tracking-uri
  • --mlflow-experiment-name
  • --mlflow-run-name
• Added mlflow dependency to requirements.txt.
• Updated README.md with MLflow usage options.
• Added/extended tests in tests/test_run_evaluation.py:
  • existing output CSV behavior remains covered
  • new test verifies MLflow params, metrics, and artifacts are logged when enabled
• Added .env.example with required Azure OpenAI variables and optional MLflow environment variables.

Why

This enables reproducible, trackable evaluation runs with experiment history and artifacts, making prompt/model comparison easier over time and aligning this repo with the evaluation observability pattern used in the referenced WIS script.

[SamuelHLewis]

@copilot review this PR, focusing on how results are logged in Azure MLFlow

[Copilot]

@SamuelHLewis I've opened a new pull request, #15, to work on those changes. Once the pull request is ready, I'll request review from you.

[SamuelHLewis]

Copilot review only flagged the possibility of making MLFlow optional, which we're deliberately not doing, so no changes are needed after this