BUG: `DataFrame[StringDtype].where(DataFrame[bool], list[str])` returns object type instead of StringDtype.

[mroeschke]

In [1]: import pandas as pd

In [2]: pdf = pd.DataFrame({"A": ["a", "bc", "cde", "fghi"]})

In [3]: pdf.dtypes
Out[3]: 
A    str
dtype: object

In [4]: pdf_mask = pd.DataFrame({"A": [True, False, True, False]})

In [6]: result = pdf.where(pdf_mask, ["cudf"])

In [7]: result
Out[7] 
      A
0     a
1  cudf
2   cde
3  cudf

In [11]: pdf.dtypes.iloc[0]
Out[11]: <StringDtype(na_value=nan)>

In [12]: result.dtypes.iloc[0]
Out[12]: dtype('O')

I would expect the result to preserve a StringDtype instead of return object