BUG: NDFrame._where returns an unexpected dtype

[ELHoussineT]

Pandas version checks

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• I have confirmed this bug exists on the main branch of pandas.

Reproducible Example

import pandas as pd 
s = pd.Series(['a','b']) # dtype: object
s.mask([True, True], 1) # dtype: object but expected to be int64

Issue Description

NDFrame._where which is used under the hood of .where and .mask returns sometimes unexpected dtype.

To illustrate, the final Series in the example below is expected to have int64 dtype but it remains a object.

>>> s = pd.Series(['a','b'])
0    a
1    b
dtype: object

>>> s.mask([True, True], 1)
0    1
1    1
dtype: object

This issue came to our attention during the construction of this PR: #50343 (comment)

Expected Behavior

It is expected for the resulting object to "refresh" its dtype. In the example above, the dtype shall be int64.

Installed Versions

INSTALLED VERSIONS

commit : 2e218d1
python : 3.10.6.final.0
python-bits : 64
OS : Linux
OS-release : 5.19.0-38-generic
Version : #39~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Fri Mar 17 21:16:15 UTC 2
machine : x86_64
processor : x86_64
byteorder : little
LC_ALL : None
LANG : en_US.UTF-8
LOCALE : en_US.UTF-8

pandas : 1.5.3
numpy : 1.23.5
pytz : 2022.7
dateutil : 2.8.2
setuptools : 65.6.3
pip : 22.3.1
Cython : 0.29.32
pytest : 7.2.0
hypothesis : 6.61.0
...
xlrd : 2.0.1
xlwt : None
zstandard : 0.19.0
tzdata : 2022.7

[rhshadrach]

I disagree, I don't think we should be doing inference here. If you instead did s.mask([True, False], 1), then this would need to be object dtype. The dtype of the result should not depend on the values of the mask.

[phofl]

Agreed with @rhshadrach