Pandas BUG: pd.NA results in wrong interpolation result

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

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

• [ ] I have confirmed this bug exists on the master branch of pandas.

Reproducible Example

>>>pd.Series([pd.NA, 1, pd.NA]).interpolate()
0    <NA>
1       1
2    <NA>
dtype: object

>>>pd.Series([np.nan, 1, np.nan]).interpolate()
0    NaN
1    1.0
2    1.0
dtype: float64

Issue Description

pd.NA results in wrong interpolation result, which is incompatible with the np.nan result.

Expected Behavior

>>>pd.Series([pd.NA, 1, pd.NA]).interpolate()
0   <NA>
1           1
2           1
dtype: Int64

Installed Versions

1.3.4

Comment From: phofl

Hm this looks correct to me. Your series has object dtype, this is consistent with

ser = pd.Series([np.nan, 1, np.nan], dtype=object).interpolate()

If you set the dtype to Int64, this raises as well, because linear is not supported there.