[EXTERNAL] docs(matrix-factorization): add measurable validation, outputs, and performance checks to subject and audit

[vpollo11]

Changes ✅ :

• Added reproducibility and metric-saving requirements (RMSE, MAE, % improvement) under reports/model_metrics.json.
• Added normalization and train/test split steps with random_state=42 for reproducibility.
• Required saved artifacts:
  • reports/svd_predictions.npy
  • reports/pmf_convergence.png
  • reports/rmse_comparison.png
  • reports/predicted_vs_actual.png
  • reports/user__recommendations.csv
• Defined performance thresholds:
  • SVD RMSE ≤ 0.90
  • PMF RMSE ≤ 0.85
  • PMF improvement ≥ 5%
• Added Performance & Validation section to subject:
  • Includes expected metrics, visualization outputs, and fail conditions.
• Updated audit with detailed checks:
  • File presence validation for models, utils, and reports.
  • Verification of saved plots, JSON metrics, and user recommendation files.
  • Runtime check for streamlit run app.py and error handling validation.
• Ensured 1:1 alignment between subject and audit with explicit measurable outcomes.

[Oumaimafisaoui]

LGTM