Micro average for auroc

[FlorianMF]

🚀 Feature

Add 'micro' to list of allowed averages.

Motivation

For now, 'average' must be in [None, 'macro', 'weighted].
For multi-label classification it would be nice to allow micro averaging.
This comes down to calculating auroc(preds.flatten(), target.flatten()).

Similarly to https://scikit-learn.org/stable/modules/generated/sklearn.metrics.roc_auc_score.html.
One could also consider adding 'samples' to the list of accepted averages.

What do you think?