Make tests fail, not print

[sneakers-the-rat]

The output of the tests is pretty difficult to read, but there were lots of places that were saying "ERROR" without failing.

Looking further, it appears as if the default behavior for the TestEnvironment is to print, not fail on error.

There are also many print statements throughout the tests - raising an exception should be the print statement in tests.

This PR

• Changes the default test behavior to fail on errors
• Raises exceptions rather than logging when tests should fail
• Removes all print statements from tests.

[codecov[bot]]

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Project coverage is 62.75%. Comparing base (33ca663) to head (31a8ad0). Report is 12 commits behind head on main.

Additional details and impacted files

```diff @@ Coverage Diff @@ ## main #312 +/- ## ========================================== - Coverage 62.92% 62.75% -0.18% ========================================== Files 62 61 -1 Lines 8545 8521 -24 Branches 2437 2437 ========================================== - Hits 5377 5347 -30 - Misses 2557 2560 +3 - Partials 611 614 +3 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[sneakers-the-rat]

The errors here are from comparisons being done to output files that don't exist - so the test is trivial, and the errors are correct, but should be conditioned on something like --generate-snapshots.

So we need to either:

• remove tests or
• add snapshots

The fact that the tests are failing on windows but the check appears to be passing in the gh interface indicates that the tests as currently configured don't actually alert us to failures (for windows). I assume that's from some weirdness in the way unittest deals with return codes, so we should switch to pytest ASAP.