[ArrowStringDtype] Make it already a StringDtype subclass

[jorisvandenbossche]

@simonjayhawkins this will be redundant after #39908, but in the meantime, ArrowStringDtype being a subclass would help for writing robust code related to StringDtype in downstream packages (for those testing against pandas master)

[simonjayhawkins]

Thanks @jorisvandenbossche lgtm

we could clean-up the various isinstance(..., StringDtype) -> isinstance(..., (StringDtype, ArrowStringDtype)) checks. do that here or as a follow-up?

[jorisvandenbossche]

we could clean-up the various isinstance(..., StringDtype) -> isinstance(..., (StringDtype, ArrowStringDtype)) checks. do that here or as a follow-up?

Did that here

[simonjayhawkins]

pandas/core/arrays/string_arrow.py:113: error: Return type "Type[ArrowStringArray]" of "construct_array_type" incompatible with return type "Type[StringArray]" in supertype "StringDtype"  [override]
pandas/core/arrays/string_arrow.py:129: error: Return type "ArrowStringArray" of "__from_arrow__" incompatible with return type "StringArray" in supertype "StringDtype"  [override]

[jorisvandenbossche]

@simonjayhawkins is there an easy solution for that? (sorry, currently don't have time to dive into this, otherwise will add some ignores)

[simonjayhawkins]

in #39908 we add ArrowStringArray to the return types and import it in a TYPE_CHECKING block

[jorisvandenbossche]

Would you be fine with ignoring it for now, and then properly fixing it in #39908 ?
I could copy the approach from #39908, but that also seems a bit strange to do right now since the StringDtype itself doesn't yet return ArrowStringArray.

[simonjayhawkins]

sure