expand probe feature selection to compare discrete variables to discrete probes and continuous variables to continuous probes

[solegalli]

maybe we could add a parameter to flag the variables that are discrete like the mutual_info_classif from sklearn

[solegalli]

I think this is better addressed by showing how to do it in the documentation, instead of trying to handle it automatically.

Add a demo where we compare discrete features against discrete probes and continuous features against continuous probes maybe even in a pipeline.

I am also thinking of making this demo on Kaggle and linking to the documentation instead of adding it to the docs, otherwise the user guide becomes too extensive, and I fear nobody will read it.