Update unit tests for allocate net gen

[jdangerx]

While talking about @grgmiller 's PR #2235, @cmgosnell and I decided to take a crack at testing the allocation behavior.

There's a few assertions that fail due to some bugs in our "allocating stuff that is missing in various places" code - I decided that fixing those bugs is out of scope of "get the tests working again."

TODO:

• we didn't convert the other xfail test yet
• we could probably come up with a cleaner looking way to provide input data
• we could probably assert other things about the allocation beyond "the sums match"

[jdangerx]

@cmgosnell I got the urge to clean these up a bit more. Take a look! We should probably do some more complicated asserts, and maybe test the other top-level function (aggregate_gen_fuel_by_generator), but I think the pattern is now easier to follow.

[codecov]

Codecov Report

Base: 86.0% // Head: 86.0% // Increases project coverage by +0.0% 🎉

Coverage data is based on head (bc1b936) compared to base (fcee14b).
Patch coverage: 100.0% of modified lines in pull request are covered.

Additional details and impacted files

@@          Coverage Diff          @@
##             dev   #2297   +/-   ##
=====================================
  Coverage   86.0%   86.0%           
=====================================
  Files         74      74           
  Lines       9273    9273           
=====================================
+ Hits        7980    7981    +1     
+ Misses      1293    1292    -1     

Impacted Files	Coverage Δ
src/pudl/analysis/allocate_net_gen.py	97.0% <100.0%> (+0.3%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

[grgmiller]

I might suggest setting the threshold lower than 5% or 7%. In theory, the allocated totals should exactly match the input totals, although in practice perhaps rounding errors might create slight differences. Maybe a stricter 1% (or 0.5%?) threshold would catch situations where the allocation is off while allowing some small tolerance for rounding? Or should we test that they exactly match?

[cmgosnell]

these changes definitely make this easier to read and navigate! Appreciate the fixture - that should make it easier to test different things. We have a long way to go to make this module less 🐛y but hopefully this will be a good place to build from.

[cmgosnell]

this is great! good use of the "don't make the full output dataframe" methodology

[cmgosnell]

👶🏻 thing but this is now an arg so you don't need it in here