Enhance data processing and visualization for neuroimaging analysis

• Improved build_dataframe function to exclude irrelevant segments using MNI brain mask.
• Added brain mask computation and segment retention based on brain voxel count.
• Refined segment id calculations using np.clip for boundary handling.
• Integrated plotting functionality for segment views in plot.py, including new plot_segment_views function.
• Updated plot_feature_importance function to return a sorted dataframe of feature importances.
• Modified train.py to handle multi-file processing with combined volumetric data.
• Implemented segment importance visualization in training pipeline with SHAP values.
• Other minor optimizations and refactorings for improved readability and functionality.