Build model validation framework and automated performance reports

[tinghf]

Create a standardized validation framework that tests calibration and prediction capabilities over time.
Implement rolling-window experiments that fit and forecast across multiple horizons and data slices.
Compute calibration and predictive metrics and store them in a structured format.
Generate standardized tables and figures for specific combinations of model and data versions.
Automate report generation via CI or scheduled jobs where feasible.