fix(python): Make to_pandas() work for Dataframe and Series with dtype Object

[itamarst]

Fixes #13021

[stinodego]

Thanks for the PR! I left a bunch of small comments.

[itamarst]

@stinodego ready for another round of review, if you have a moment.

[stinodego]

A few more comments. With that, this should be good to merge.

I am not too happy with splitting up the implementation up like this (different strategies for different columns, then merge them). The Object columns are not converted in parallel and I am not sure about the performance implications of inserting them one-by-one into a pandas dataframe (maybe this is free, not sure).

For now it seems warranted to correctly handle Object types. But I wonder what a more ideal solution would look like.

PS: Could you send me another message on Discord? I lost our previous conversation and I can't find you by your username.

[itamarst]

@stinodego back to you.

[ritchie46]

Thank you @itamarst