Improved summaries: refactor summary code out of generated jupyter notebooks into MMEDs and improve summary code overall

[mstapylton]

Is your feature request related to a problem? Please describe.
errors keep coming up in the summaries themselves rather than the code that generates them, even though the issue originates in the code that generates the summaries.

An example of this is when we get errors that look like Rcolorbrewer issues from the jupyter conda environment, but are really with what we're passing to that code as unique values.

Describe the solution you'd like
The code that generates summaries should be cleaned up and code that runs in the summaries refactored to MMEDs where possible. The idea is to make the summaries themselves run as little code as possible, to catch more issues further upstream and have better seperation between MMEDs and the tools MMEDs runs.