Index arithmetic overloads show wrong types

[MarcoGorelli]

Describe the bug

I think this is an issue:

pandas-stubs/pandas-stubs/core/arraylike.pyi

Line 24 in fdda5a6

def __sub__(self, other: Any) -> Self: ...

as the dtype isn't preserved

To Reproduce

1. Provide a minimal runnable pandas example that is not properly checked by the stubs.
2. Indicate which type checker you are using (mypy or pyright).
3. Show the error message received from that type checker while checking your example.

from __future__ import annotations
import pandas as pd
from typing import reveal_type

idx = pd.Index([1,2,3])

reveal_type(idx - 1.5)

at runtime this gives

Index([-0.5, 0.5, 1.5], dtype='float64')

but it's revealed to be Index[int] by type checkers

Please complete the following information:

• OS: [e.g. Windows, Linux, MacOS]
• OS Version [e.g. 22]
• python version
• version of type checker
• version of installed pandas-stubs 3033eea

Additional context
Add any other context about the problem here.

[Dr-Irv]

Yes, there's been a lot of recent (and still ongoing) work to fix up the arithmetic for Series, and probably something similar should be done for Index because Index is using what is in pandas-stubs\core\arraylike.pyi, and needs special handling, by putting methods into the Index class and its subclasses.

[MarcoGorelli]

Once this is addressed, it should be possible to uncomment some of the failing test cases in #1348 (review)