fix(python): Verify the integrity of pandas column names before implied string conversion

[tylerriccio33]

In the issue #16025 , the author converts a pandas dataframe who's columns are 0 and '0'. In polars dataframes, this is illegal, as all the column names are converted to strings. This is done in the pl.from_pandas, but the conversion happens in a way that overwrites a prior column. For example, if 0 and '0' are columns, the '0' will overwrite the first column, since that column was converted first, to a '0'.

I check the uniqueness of the stringified names in from_pandas. This should also catch some unusual column names, like arbitrary objects with str methods.

I did change the behavior of test_from_pandas_duplicated_columns in test_interop. The test now raises the error message I wrote since the duplicate columns were caught earlier. I understand this may be undesirable since it's more consistent to propagate the pandas error up the stack to the user, which was the original behavior.

Also, I opened a pull request on this issue yesterday but deleted it since I was on the wrong branch. Excuse my git skills, I'm a beginner with it. Let me know if you need any changes to this or have any questions! Thank you.

[stinodego]

Please rebase your branch on the main branch, then we can review this.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 80.45%. Comparing base (200c6a4) to head (2f15ea2).

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #17433      +/-   ##
==========================================
- Coverage   80.46%   80.45%   -0.01%     
==========================================
  Files        1483     1483              
  Lines      194832   194836       +4     
  Branches     2770     2771       +1     
==========================================
- Hits       156767   156752      -15     
- Misses      37556    37574      +18     
- Partials      509      510       +1     

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

[tylerriccio33]

Please rebase your branch on the main branch, then we can review this.

I rebased and re-pushed the pr, which triggered the ci checks. Let me know if you need anything else!

[ritchie46]

Thanks @tylerriccio33