TYP: Non-overlapping identity checks with pd.NaT

[twoertwein]

Either wrong type annotations in pandas/core/ops/mask_ops.py or there are a few branches that will never get hit (bool | np.ndarray can never be NAType).

Found in #43744

pandas/core/ops/mask_ops.py:49: error: Non-overlapping identity check (left operand type: "Union[bool, ndarray[Any, Any]]", right operand type: "NAType") [comparison-overlap]
pandas/core/ops/mask_ops.py:66: error: Non-overlapping identity check (left operand type: "Union[bool, ndarray[Any, Any]]", right operand type: "NAType") [comparison-overlap]
pandas/core/ops/mask_ops.py:106: error: Non-overlapping identity check (left operand type: "Union[bool, ndarray[Any, Any]]", right operand type: "NAType") [comparison-overlap]
pandas/core/ops/mask_ops.py:114: error: Non-overlapping identity check (left operand type: "Union[bool, ndarray[Any, Any]]", right operand type: "NAType") [comparison-overlap]

edit: happens in a few more places as well

pandas/core/arrays/timedeltas.py:537: error: Non-overlapping identity check (left operand type: "Timedelta", right operand type: "NaTType") [comparison-overlap]
pandas/core/arrays/timedeltas.py:594: error: Non-overlapping identity check (left operand type: "Timedelta", right operand type: "NaTType") [comparison-overlap]
pandas/core/arrays/timedeltas.py:637: error: Non-overlapping identity check (left operand type: "Timedelta", right operand type: "NaTType") [comparison-overlap]
pandas/core/arrays/timedeltas.py:705: error: Non-overlapping identity check (left operand type: "Timedelta", right operand type: "NaTType") [comparison-overlap]

pandas/core/arrays/masked.py:341: error: Non-overlapping identity check (left operand type: "Union[Union[str, int, float, bool], Union[Period, Timestamp, Timedelta, Interval]]", right operand type: "NoDefault") [comparison-overlap]
pandas/core/arrays/masked.py:349: error: Non-overlapping identity check (left operand type: "Union[Union[str, int, float, bool], Union[Period, Timestamp, Timedelta, Interval]]", right operand type: "NAType") [comparison-overlap]

[jbrockmendel]

with the Timedelta one the trouble is that Timedelta.__new__ can return pd.NaT, which is in the timedeltas.pyi file, but mypy complains about so it has a 'type: ignore'

[twoertwein]

Based on codecov, all except these two lines are hit:

pandas/core/arrays/timedeltas.py:637: error: Non-overlapping identity check (left operand type: "Timedelta", right operand type: "NaTType") [comparison-overlap]
pandas/core/arrays/timedeltas.py:705: error: Non-overlapping identity check (left operand type: "Timedelta", right operand type: "NaTType") [comparison-overlap]

I guess this suggests that the type annotations are incorrect in most of the above places.

[twoertwein]

Closing this as these are just type issues that can in the simplest case be "solved" by type: ignore.