BUG: invalidate cached Index._isnan if use_inf_as_na changed?

[jbrockmendel]

• 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 master branch of pandas.

Reproducible Example

import numpy as np
import pandas as pd

idx = pd.Index([np.inf, np.nan], dtype=np.float64)

assert (pd.isna(idx) == idx.isna()).all()

pd.set_option("use_inf_as_na", True)

assert (pd.isna(idx) == idx.isna()).all()  # <- nope!

Issue Description

pd.isna(obj) and obj.isna() are generally interchangeable, but pd.Index caches .isna in a _isnan cache_readonly. If we want to maintain that interchangeability, we'd need to invalidate that cache.

My inclination is that it isn't worth futzing around with, but someone who uses use_inf_as_na=True might care more.

Expected Behavior

assert (pd.isna(idx) == idx.isna()).all()

Installed Versions

Replace this line with the output of pd.show_versions()

[mroeschke]

Just a heads up: some of your issues have examples have an extra ``` at the bottom of the formatting block which messes up the rendering of the issue

[jbrockmendel]

Thanks; still getting used to the new templates.

[simonjayhawkins]

removing milestone

[jbrockmendel]

#53494 deprecated use_inf_as_na, so closing.