DEPR/API: combine

[jbrockmendel]

ser = pd.Series([1, 2, None], dtype="int64[pyarrow]")
ser2 = pd.Series([None, 2, 1], dtype="int64[pyarrow]")
df = ser.to_frame()
df2 = ser2.to_frame()

>>> ser.combine(ser2, operator.add, fill_value=-1)
0    <NA>
1       4
2    <NA>
dtype: int64[pyarrow]
>>> df.combine(df2, operator.add, fill_value=-1)
   0
0  0
1  4
2  0

These behave differently for Series and DataFrame in a way that I find not-at-all intuitive. We should either align or deprecate these.

[tpackard1]

Hi I'd like to take a stab at this issue. But is there a need to discuss whether the they should be aligned or deprecated? I'm going to go ahead and do a take but if discussion is needed before a PR is submitted please let me know.

[tpackard1]

take

[lithomas1]

@tpackard1 This needs some discussion on whether the entire method should be deprecated or the behaviors aligned.

I'd recommend looking at some other issues for now.

[jreback]

+1 to deprecate / remove.