refactor: Discontinue most parts of mypy plugin

[delsner]

Motivation

Closes #54.

Changes

Discontinue mypy plugin features that are hard to maintain and provide limited value.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (fdd8977) to head (0f4ad8e).
⚠️ Report is 7 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff            @@
##              main      #103   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           41        40    -1     
  Lines         2401      2389   -12     
=========================================
- Hits          2401      2389   -12     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[borchero]

That's a lot of complexity gone 😅

[runkelcorey]

I'm a little confused on the scope of this change. Will mypy still be able to validate some rules of some collections?

[borchero]

Sorry for the confusion @runkelcorey! All the features that you'd expect from mypy will still work when this is merged.

First of all, this change only has an effect if you use dataframely's mypy plugin. Even then, the only things that change are the following:

• Calling iter_rows(named=True) on a dy.DataFrame does not return an iterator of TypedDicts with the correct keys and values anymore but a dict[str, Any], just like polars.
• When filtering a collection, the returned failure info dictionary is not a TypedDict anymore but a dict[str, dy.FailureInfo].

Both of these things did more harm than good. The former was sometimes evaluated incorrectly when mypy used caching. The latter caused the user to not be able to use the failure info dictionary as dict[str, dy.FailureInfo] (because a TypedDict is not interpreted as such). Hence, we decided to remove these two things (which also caused plenty of complexity).

So tl;dr: if you do not rely on these two things, this PR won't have an effect on your usage.