handle pyright inferring mixed float and int lists

[Dr-Irv]

pyright with strict type checking is inferring that a list of mixed int and float is list[int | float], so created overloads for Series.__new__() that makes that a Series[float].

[twoertwein]

It would be nice if we could re-write all the overloads more generically, but I assume the union issue (since we use TypeVar with bounds) and the Series sub-classes prevent us from using a more generic approach

    def __new__(
        cls,
        data: Sequence[S1],
        index: Axes | None = ...,
        *,
        dtype: type[S1] = ...,
        name: Hashable = ...,
        copy: bool = ...,
    ) -> Series[S1]: ...

[Dr-Irv]

It would be nice if we could re-write all the overloads more generically, but I assume the union issue (since we use TypeVar with bounds) and the Series sub-classes prevent us from using a more generic approach

Yes, that was exactly the problem. We actually have this overload in there already:

    @overload
    def __new__(
        cls,
        data: S1 | _ListLike[S1] | dict[Hashable, S1] | dict_keys[S1, Any],
        index: Axes | None = ...,
        *,
        dtype: Dtype = ...,
        name: Hashable = ...,
        copy: bool = ...,
    ) -> Self: ...

where _ListLike is defined as:

_ListLike: TypeAlias = (
    ArrayLike | dict[_str, np.ndarray] | Sequence[S1] | IndexOpsMixin[S1]
)

The issue that this PR fixes relates to the union problem of int | float

@twoertwein will you also merge?