Fix tween_method() type validation

[KoBeWi]

Fixes #72808

tween_method() will now validate types and auto-convert between int and float, just like tween_property().

[TokageItLab]

I think this PR is fine assuming the current implementation.

BTW, what do you think about #53548? Maybe we should discuss whether double should be cast int first and interpolated as in current, or whether all int values should be double once and cast last.

[akien-mga]

Thanks!

[KoBeWi]

@TokageItLab Not sure what you mean. I think float/int interpolation is fine. I left a comment in the PR.

[TokageItLab]

@KoBeWi For example, if you want to transition from 1 to 3.5, int(1)->int(3.5) and int(double(1.0)->double(3.5)) would give different results. Which is ideal?

[KoBeWi]

Good question.
I didn't really see anyone complain about it, so I'd leave it as is for now. Depending on the feedback we might add a warning or rework as you suggest.
I think the most common use-case when mixing ints and floats is either when float is a rounded value or when int is the final value. In both cases the conversion is irrelevant.

[TokageItLab]

Well, I agree that it is certainly not that common a case and is in the category of warnings and documentation issues.