IoU probe does not work for multi-class problems

[fdalvi]

NeuroX/neurox/interpretation/iou_probe.py

Line 37 in f2314cc

scores.append(average_precision_score(y_train, X_train[:, i]))

assumes that y_train only has 0's and 1's. For multiclass, we should run IoU for each tag, and then combine into a global ranking.