Consolidate pre-commit logic

[camposandro]

This pull request consolidates the code quality checks, removing the dedicated code style analysis pipeline and using pre-commit to handle the checks both locally and on the CI. Closes #410.

• Removes the code_style.yml.jinja script
• Adjusts pre-commit.ci to run over all files, not just changed files
• Adjusts pre-commit.ci to disable fail-fast so all code style issues are surfaced at once
• Makes the unit test hook run last because it's the most expensive

If configured, a bot will push automatic fixes to non-draft PRs to apply code formatting, ordering of imports and clear notebook outputs, if there are any (example). Most linting issues need to be addressed manually by the user.

Checklist

• This PR is meant for the lincc-frameworks/python-project-template repo and not a downstream one instead.
• This change is linked to an open issue
• This change includes integration testing, or is small enough to be covered by existing tests

[drewoldag]

Looks good to me. Thanks!

[drewoldag]

Still looking good. Thanks for the extra work!

[delucchi-cmu]

Can we leave the fail_fast behavior in, for running locally, but take it out in the CI? Or could I enable it locally by default? I find it useful to not run the unit tests if I know I'm going to have to change something and just run everything all over again.

[camposandro]

Unfortunately there seems to be no way of running pre-commit with a fail_fast flag. We could potentially create two separate pre-commit-config, one for the CI and another for the local runs. Might be overkill but we could remove all the hooks we're actually not using on the CI and make the configuration more clear to the users. What is your opinion on this? Tagging @drewoldag too.