Inconsistent overload resolution on `Generic` methods

[Dr-Irv]

Describe the bug

According to the pyright docs at https://github.com/microsoft/pyright/blob/main/docs/type-inference.md and the typing docs at https://typing.python.org/en/latest/spec/overload.html#step-5, if there is a conflict in return types of matching overloads, then pyright should return Unknown. In the example below, that is not happening. mypy is reporting Any, which we believe is what should happen, while pyright is returning Never.

Code or Screenshots
PYI file gensub.pyi:

from __future__ import annotations
from typing import (
    Any,
    overload,
    Generic,
    Never,
    reveal_type,
    Self,
)
from typing_extensions import TypeVar

T = TypeVar("T", int, str)

class Se(Generic[T]):
    @overload
    def __sub__(self: Se[int], other: Se[int]) -> Never: ...
    @overload
    def __sub__(self, other: Self) -> Self: ...

PY file gensubtst.py:

from typing import Any
from gensub import Se


def t1() -> None:
    reveal_type(Se[Any]() - Se[Any]())  # mypy: Any, pyright: Never.

pyright gensub.py output:

gensubtst.py:2:6 - warning: Import "gensub" could not be resolved from source (reportMissingModuleSource)
gensubtst.py:6:17 - information: Type of "Se[Any]() - Se[Any]()" is "Never"

mypy gensub.py output: (version 1.17.1)

gensubtst.py:6: note: Revealed type is "Any"

We think mypy is correct here, because the result should be Unknown for mypy since both overloads match. This is coming up in a PR we are working on for pandas-stubs.

VS Code extension or command-line
pyright command line 1.1.405, python 3.11

[erictraut]

I think this is the same underlying problem as #10232, but I'll leave them both open until I confirm.