Bugfixes

[ToastyDom]

Bufixes

• rsa.py: argmin → argmax when selecting best layer in compare_model
• rsa.py: significance threshold p < 0.5 → p < 0.05 in compare_model
• searchlight.py: missing enumerate() in evaluate() loop over self.model_rdms
• searchlight.py: check_squareform called on a list instead of the numpy array inside it
• encoding.py: single-fold branch uses stale r_lst from last loop iteration instead of current layer/ROI correlations
• encoding.py: train_Ridgeregression_per_ROI saves unfitted reg instead of best_model when save_model=True
• stacked_encoding.py: identical code in both branches of the p > n vs p <= n conditional in feat_ridge_CV
• variance_partitioning_analysis.py: evaluate_4 missing triple-shared variance keys (y123, y124, y134, y234)
• variance_partitioning_analysis.py: evaluate_4 description labels bugix