DecimalTuple definition is incorrect

[dapper91]

Current DecimalTuple definition is:

class DecimalTuple(NamedTuple):
    sign: int
    digits: tuple[int, ...]
    exponent: int

source: https://github.com/python/typeshed/blob/main/stdlib/_decimal.pyi#L16

But if a decimal value is NaN the exponent is of type str. The following example illustrate that:

>>> from decimal import Decimal
>>> exponent = Decimal('NaN').as_tuple().exponent
>>> print(exponent)
n
>>> print(type(exponent))
<class 'str'>
>>>

It seems like the correct definition should be like this:

class DecimalTuple(NamedTuple):
    sign: int
    digits: tuple[int, ...]
    exponent: Union[int, str]

[JelleZijlstra]

Looks like it can also be N for a signaling NaN and F for infinity. I suppose we could do int | Literal["n", "N", "F"].