Zenith-angle dependent efficiencies

[GernotMaier]

No description provided.

[Copilot]

Pull request overview

This PR adds zenith-angle (per-zenith-bin) efficiency outputs to the classification training pipeline and updates the plotting script to visualize those zenith-binned efficiencies (with TMVA overlays becoming optional).

Changes:

• Extend evaluation_efficiency() to optionally compute and return per-zenith efficiency curves and persist them as efficiency_zeN in trained model configs.
• Update the performance-metrics plotting script to select zenith bins from joblib outputs, handle missing TMVA ROOT overlays gracefully, and write per-bin plots.
• Adjust the SHAP caching test to match the updated evaluation_efficiency() return shape.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 6 comments.

File	Description
tests/test_train_classification_shap.py	Updates the mocked evaluation_efficiency() return value to a (df, dict) tuple.
src/eventdisplay_ml/scripts/plot_classification_performance_metrics.py	Adds zenith-bin discovery/selection and optional TMVA overlays; generates per-zenith output plots.
src/eventdisplay_ml/models.py	Persists overall and per-zenith efficiency DataFrames into the model config (efficiency, efficiency_zeN).
src/eventdisplay_ml/evaluate.py	Refactors efficiency computation into a helper and adds optional per-zenith efficiency calculation/return.