Confusion matrix as a metric

[olliethomas]

Confusion matrix values are often used in metrics, but some people might want these values recording directly.

A user story would be something like
Given: everything is installed
When: I run evaluate_models and pass em.CM as a metric in the per_sens metrics
Then: a column recording the TP, FN etc counts is recorded in the results dataframe with one column per value

The bit that makes this non-trivial is that adding >1 column at a time to the results