BUG: Regex parameter from replace function doesn't work on string dtype

[GYHHAHA]

import pandas as pd
print(pd.version)
1.1.1
s1 = pd.Series(['a','b'], dtype='string')
s2 = pd.Series(['a','b'], dtype='object')
s1.replace(r'[a]',pd.NA,regex=True)
0 a
1 b
dtype: string
s2.replace(r'[a]',pd.NA,regex=True)
0
1 b
dtype: object

Hi, it seems that regex doesn't function well for the replace method on string dtype, but normal on object dtype.

Besides, I also find setting pd.NA as repl of str.replace method will raise error, both on string and object.
Yet I think it may be better not to raise this error because the pd.NA indeed could be a missing string value.

pd.Series(['a','b'],dtype='string').str.replace('a',pd.NA)
TypeError: repl must be a string or callable
pd.Series(['a','b'],dtype='object').str.replace('a',pd.NA)
TypeError: repl must be a string or callable

Thanks!

[GYHHAHA]

Oh, I find #35977 and #36038 have already discussed this bug.
Just see the second part for pd.NA permission in str.replace method.