Implement pd.NA checking function

[dsaxton]

• tests added / passed
• passes black pandas
• passes git diff upstream/master -u -- "*.py" | flake8 --diff
• whatsnew entry

This may already exist in some form, but if not I think it could be useful to have a fast function for checking specifically for the presence of pd.NA (e.g. if we're looking at an arbitrary array and don't know if we'll have access to a _mask attribute, or if we're handling some other data type that uses pd.NA for missing values but doesn't have this attribute).

[jreback]

isna already does this

[dsaxton]

isna already does this

What I was thinking here was a way to get a mask for only pd.NA (so not including np.NaN, None or other missing value indicators like isna does). The main motivation was trying to think of a way to fix #31990 without writing the equivalent of a Python for loop to do this check

[jbrockmendel]

not a big deal, but is this restriction necessary?

[dsaxton]

Maybe not, the main application I had in mind was when operating on single columns though

[jbrockmendel]

related helper function ive wanted is is_matching_na

[WillAyd]

What I was thinking here was a way to get a mask for only pd.NA (so not including np.NaN, None or other missing value indicators like isna does). The main motivation was trying to think of a way to fix #31990 without writing the equivalent of a Python for loop to do this check

Can't you use the mask that the StringDtype has to check this?

[dsaxton]

Closing since it looks like others aren't in favor of the idea. I do think some kind of "dtype agnostic" way of checking for NA could be helpful later