feat(interpret): add TreeSHAP for fast exact SHAP values on tree models

[yasmramos]

• Implement TreeSHAP algorithm for DecisionTree, RandomForest classifiers and regressors
• Add getTrees() public method to RandomForestClassifier and RandomForestRegressor
• Create TreeSHAPExample demonstrating usage with Iris dataset
• TreeSHAP provides O(T*L²) complexity vs O(2^M) for Kernel SHAP
• Support for explain(), explainBatch(), and meanAbsoluteShap() methods
• Include expected value computation and SHAP value verification

This enables production-ready model interpretability for tree-based models
with significant performance improvements over model-agnostic approaches.