GH-34449 [Python] array(pd.Categorical) raising for arrow-backed cat

[phofl]

Rationale for this change

This occurs when saving pandas arrow-backed categoricals to parquet. There might be a better way to do this than calling combine_chunks, but couldn't find a way to convert a ChunkedArray to a DictionaryArray

What changes are included in this PR?

Bugfix and tests only

Are these changes tested?

Yes

Are there any user-facing changes?

Yes, bug fix

• closes [Python] to_parquet fails with a category field backed by pyarrow string #34449

• Closes: [Python] to_parquet fails with a category field backed by pyarrow string #34449

[github-actions]

• Closes: [Python] to_parquet fails with a category field backed by pyarrow string #34449

[github-actions]

⚠️ GitHub issue #34449 has been automatically assigned in GitHub to PR creator.

[AlenkaF]

Thank you @phofl for contributing!

As I mentioned in the issue comments, there is already a PR for a duplicated GitHub issue: #34289

As you can see in the comments of the duplicated PR issue #33727 (comment), the failure happens because pa.DictionaryArray.from_array doesn't accept chunked arrays. In this case, the ChunkedArray will always have only one chunk and so a better way of solving the issue is to return an array in case of single chunked data when calling _handle_arrow_array_protocol (__arrow_array__).

[phofl]

Sorry I should have checked the arrow issue. I came here through debugging the pandas equivalent. Closing then, sorry for the overhead

[AlenkaF]

No problem at all! Thank you for making an effort 🙏