Add callable condition type in Series.mask, DataFrame.mask

[georgesittas]

• Closes Passing a callable to DataFrame.mask's 'cond' argument results in arg-type error (mypy) #917
• Tests added: Please use assert_type() to assert the type of any return value

I played around a bit and found that Series.mask() actually had the same issue as DataFrame.mask():

import pandas as pd
s = pd.Series([1]).mask(cond=lambda x: x % 2 == 1, other=2)
print(s)
# 0    2
# dtype: int64

(.venv) ➜  test python -m mypy test.py
test.py:2: error: Argument "cond" to "mask" of "Series" has incompatible type "Callable[[Any], Any]"; expected "Series[bool] | ndarray[Any, dtype[bool_]] | list[bool]"  [arg-type]
Found 1 error in 1 file (checked 1 source file)

I tried to fix both in this PR. The new cond types match those used in Series.where and DataFrame.where. Let me know if it's desirable to factor the cond type out by moving it to _typing.pyi.

[georgesittas]

CI failed with:

error: Unnecessary "# pyright: ignore" rule: "reportIncompatibleMethodOverride"

I saw this one locally as well, but wasn't sure if it was due to my configuration. Should we get rid of that rule from the ignore hint? The test suite ran fine for me locally after removing it, but haven't investigated what it does / implications of removing it.

[Dr-Irv]

CI failed with:

error: Unnecessary "# pyright: ignore" rule: "reportIncompatibleMethodOverride"

I saw this one locally as well, but wasn't sure if it was due to my configuration. Should we get rid of that rule from the ignore hint? The test suite ran fine for me locally after removing it, but haven't investigated what it does / implications of removing it.

I am guessing that pyright has a new version that says that the rule is unnecessary, so you can remove it as part of this PR. It won't have an effect - we put those in there when we disagree with the type checker, so in this case, they fixed a bug making the ignore statement unnecessary.

[georgesittas]

Got it - done.

[Dr-Irv]

Thanks @georgesittas