Allow unary + in Literal integers

[dmoisset]

As discussed in https://discuss.python.org/t/signed-values-in-literal-type/39919

[JelleZijlstra]

Looks good, but it would be good to get an implementation in at least one type checker. It should be trivial, but as a general principle I'd like the spec to only specify behavior that has actually been implemented.

While reviewing the change I was thinking about whether we should specify what + means. My conclusion is that it's fine to leave it implicit that it means the same thing as it means at runtime.

[JelleZijlstra]

I wrote an implementation for mypy, which was trivial as expected: python/mypy#16729.

[erictraut]

Oh, one thing I just thought about. Should a type checker explicitly reject redundant unary + or - operators in a literal type? Example: Literal[+++++-----1]. Do we care one way or another?

[JelleZijlstra]

Good point. I'd say any number of +/- should be accepted as it's unambiguous, easy to implement, and matches the runtime behavior.

[rchen152]

Looks good to me. I think it's fine to leave the edge case of whether multiple +/- are allowed unspecified.