REGR: Fix to_numpy for masked array with non-numeric dtype

[phofl]

• closes BUG: to_numpy_dtype_inference returns None when array is non-numeric and dtype isn't given #56991 (Replace xxxx with the GitHub issue number)
• Tests added and passed if fixing a bug or adding a new feature
• All code checks passed.
• Added type annotations to new arguments/methods/functions.
• Added an entry in the latest doc/source/whatsnew/vX.X.X.rst file if fixing a bug or adding a new feature.

[mroeschke]

Needs a whatsnew otherwise LGTM

[lukemanley]

sorry to ask this here, but its somewhat related. Should the following return pd.NA or np.nan for the na_value? It is returning pd.NA but that isn't allowed in a float64 array:

In [1]: import pandas as pd

In [2]: from pandas.core.arrays._utils import to_numpy_dtype_inference

In [3]: from pandas._libs import lib

In [4]: arr = pd.array([1, pd.NA], dtype="int64[pyarrow]")

In [5]: to_numpy_dtype_inference(arr, dtype="float64", na_value=lib.no_default, hasna=True)
Out[5]: (dtype('float64'), <NA>)

[mroeschke]

Should the following return pd.NA or np.nan for the na_value? It is returning pd.NA but that isn't allowed in a float64 array:

Personally I would expected np.nan here

[phofl]

#57233

[phofl]

@mroeschke good to merge?

[mroeschke]

Thanks @phofl