Allow to generate scatter plots/confusion matrices

[MariusMerkleQC]

From @borchero:

This is still a very open idea: when comparing pipeline outputs, we are often interested in how model predictions/scores change. To this end, we often generate scatter plots/confusion matrices.

Potentially, we could support this to some extent via diffly?

[MariusMerkleQC]

From @EgeKaraismailogluQC:

To make this idea more concrete: I can think of adding an instance method to DataFrameComparsion that accepts a column name and displays either a scatter plot if the column is real-valued or a confusion matrix if it is categorical. Is this similar to what you had in mind?

[MariusMerkleQC]

From @borchero:

Yes! You could also consider generating multiple plots, one for each common column, if no column is provided as parameter.