bugfix/1835: Keep nulls in polars when dropping invalid rows and nullable=True

[baldwinj30]

When nullable=True in polars, it seems the check_outputs values from the error handler are left as null rather than True where values are null. This would be one possible way to remedy that to match with pandas behavior as per #1835 however I am not sure if this is the best place to handle that behavior, or even if it makes sense to match pandas behavior here or not.

There also seemed to be a bug in the original test for polars dropping null values which I attempted to fix up here.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 93.32%. Comparing base (812b2a8) to head (17a1967).
Report is 186 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1890      +/-   ##
==========================================
- Coverage   94.28%   93.32%   -0.96%     
==========================================
  Files          91      121      +30     
  Lines        7013     9301    +2288     
==========================================
+ Hits         6612     8680    +2068     
- Misses        401      621     +220     

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

[cosmicBboy]

thanks for the contribution @baldwinj30! I do think it makes sense to match pandas behavior here, but the kwarg of interest is Check(ignore_na=True). This is handled in the check backend layer in the postprocess method, see the pandas implementation:

https://github.com/unionai-oss/pandera/blob/main/pandera/backends/pandas/checks.py#L273

The equivalent spot in the polars check backend is here: https://github.com/unionai-oss/pandera/blob/main/pandera/backends/polars/checks.py#L90, let's make the change over there

[baldwinj30]

Thanks @cosmicBboy! I moved the fix in to the check backend. I am not totally sure about the failing checks, but I don't think they are related to my changes?

[cosmicBboy]

thanks @baldwinj30 ! 🚀