Pandas (as of 1.3.0) no longer supports passing `norm` - used in evaluation.py.

[benjeffery]

This is an issue the 1.2 series neither has wheels or is marked in conda as compatible with 3.10. So we need pandas>=1.3.0 to run CI on python 3.10. This maybe an issue with pandas, where by it should be allowing norm to pass through if specified by the caller rather than specifying it itself leading to TypeError: matplotlib.axes._axes.Axes.scatter() got multiple values for keyword argument 'norm'. This happens at https://github.com/pandas-dev/pandas/blob/main/pandas/plotting/_matplotlib/core.py#L1102

I've commented out norm in #668 as a temporary work around to get CI going again, all in evaluation.py marked with FIXME #669.

[jeromekelleher]

Stuff in evaluation.py is basically developer utilities, so we don't need to worry too much.