Use update instead of assignment to set _charges

[ntBre]

Description

Minor update to #1066 to include _charges in the model definition. I still don't know why this is required, but it seems to work in my local tests. I just realized that it might be necessary to do self._charges.clear() before self._charges.update() to avoid mixing charges. I read the if condition as len == 0 and ... instead of or, which could possibly allow update to be called on a dict already containing some charges.

Checklist

• Add tests - should be covered by existing tests
• Lint
• Update docstrings - n/a

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 93.58%. Comparing base (360b60c) to head (1f9608e).
Report is 2 commits behind head on develop.

Additional details and impacted files

[mattwthompson]

This ... makes a fair amount of sense to me, actually. I think in non-Pydantic world it's best practice to update a dict instead of re-assign it, but I'm also pretty sure something in Pydantic's magic machinery means just passing the return value of another function to an instance attribute gummed something up something.

[mattwthompson]

Thanks!

[mattwthompson]

I just realized that it might be necessary to do self._charges.clear() before self._charges.update() to avoid mixing charges.

For posterity, this was a Cassandra level of prescient. For some reason, no tests that existed before this PR were failing, but some tests added elsewhere (#1070) did break without clearing the cache at that point.