Standard library `_operator` module is poorly typed

[jonathandung]

The annotations for functions in the _operator module are inconsistent - some have annotations but most don't.

At first glance, most of the protocols necessary to type these incomplete signatures can be imported from _typeshed, and there is already a precedent in the stub in question.

I can work on the annotations to bridge the inconsistency, but it would be a big undertaking, so I need to get approval first.

[jonathandung]

Partially overlaps with #15601.

[hunterhogan]

FYI: #15598, which was preceded by #15577, both of which have some discussions.

[jonathandung]

Thanks

[jonathandung]

I will start working on this now. However, I am unfamiliar with the procedure of introducing new tests.

[hunterhogan]

That brought back some memories. I am familiar with making tests that don't work, and I'm very good at annoying the hell out of everyone. I'm sure you'll have a smoother experience. 😄

[jonathandung]

Also, the _SupportsDunderLT, _SupportsDunderGT, _SupportsDunderLE, _SupportsDunderLT are not necessary. These can be imported from _typeshed:

class SupportsDunderLT(Protocol[_T_contra]):
    def __lt__(self, other: _T_contra, /) -> SupportsBool: ...

class SupportsDunderGT(Protocol[_T_contra]):
    def __gt__(self, other: _T_contra, /) -> SupportsBool: ...

class SupportsDunderLE(Protocol[_T_contra]):
    def __le__(self, other: _T_contra, /) -> SupportsBool: ...

class SupportsDunderGE(Protocol[_T_contra]):
    def __ge__(self, other: _T_contra, /) -> SupportsBool: ...

Basically everything besides NotImplemented is assignable to SupportsBool, so that _SupportsDunderLT, for example, can be replaced with SupportsDunderLT[Any].

[jonathandung]

This comment that supposedly explains the presence of those redundant protocols is covered by this fix:

# The following protocols return "Any" instead of bool, since the comparison
# operators can be overloaded to return an arbitrary object. For example,
# the numpy.array comparison dunders return another numpy.array.