maint: refactor tests, fix upgrade tests (now correctly failing)

[consideRatio]

I have various soft goals I'm working towards

1. Make the tests and the infrastructure to run them a bit easier to understand and maintain
2. Make run tests emit easier to understand results if they succeed or fail Closes #899
3. Make tests runs a bit faster to complete if possible
4. Capture more test coverage if possible

I'm still deliberating on copying files to the container vs mounting the local git repo in the container. I figure maybe we can get test coverage by mounting instead of copying, but its also coupling things a bit.

Important insights

1. Our upgrade didn't work before, now they do. We have a real test failure though, and I don't aim to solve it as part of this PR. It relates to conda/mamba and tsting the tljh_extra_user_conda_packages plugin, which indicates there is an issue in the conda/mamba state after the upgrade to a new conda/mamba version.

Misc things not done

• [ ] Let bootstrap.py logs and tljh.installer logs clarify when the bootstrap.py script stops running and the tljh.installer starts running
• [ ] Stop copying files to integration test container, and mount things instead to get test coverage?
• [ ] Fix test failure discovered by fixed upgrade test
• [ ] Let our test suite run with JupyterHub.log_level set to debug

[consideRatio]

I've invested a lot of time into this PR and I could keep going because there are more things to do, but I'd love to get this merged and go from there with smaller increments going onwards.

The two test failures are real, but not caused by this PR besides fixing the tests to not be incorrectly passing as before. I've opened https://github.com/jupyterhub/the-littlest-jupyterhub/issues/918 to represent this.

[minrk]

This is wonderful! Happy to merge and move on.

[consideRatio]

Thank you @minrk!!!!!