status: add automated test suite focused on energy calculations

[t-kramer]

We had zero automated tests for src/energy.py, the core calculatiom, and were relying on a hand-checked spreadsheet after every change. This PR adds 88 tests across three tiers (fast unit tests on commit, regression snapshots on every push, full 8760h smoke tests before deployment).

For context on the design decisions and how to work with the tests day-to-day, see notes/testing-scheme.md. That's the right place to start rather than the test files themselves.

[urwahah]

thanks so much for all this work!! it looks super comprehensive. i have a few questions:

1. i'm not sure i understand the emissions rate and physics invariant tests in the tier 1 test_energy_unit.py. they aren't checking anything in energy.py - all these tests confirm is that the formulas provided in the tests are correct. if there was somehow an error in these particular calculations in the main code, i don't get how this would catch it?
2. there is a test to check that negative load values are allowed in test_loads.py. i don't know if that should be allowed? i believe the way our HR calculations are set up, the HR served load would just go to 0 if it encounters a negative load number (because of the minimum turndown), but regardless i don't know where a negative load value would be an acceptable input
3. should we have at least one test case building with measured data instead of simulated?
4. i'm unfamiliar with the testing setup so maybe an easy solution to this.. what's a good way of reviewing the differences in the regression snapshots once a change is made? i tested out the process of updating the snapshot and running git diff but that returns just an illegible output: