Revert "Allow parameters passed to GDScript functions to be nulled"

[ttencate]

This reverts commit f0efc75.

Fixes #37312.

[akien-mga]

By reverting #26528, you'll reintroduce #26047, no?

[ttencate]

@akien-mga No, the error messages have improved considerably since then. See #37312 (comment).

Edit: bother, for that particular test project posted on #26047, the error message is still the same confusing thing. Please hold while I investigate...

[ttencate]

I'm guessing it's something to do with signals vs. function calls, but I can't for the life of me figure out how to hook up signals in the 4.0 way despite having read Juan's update. None of these work:

signal my_signal

my_signal.connect(self.handler) # Nope, the identifier "my_signal" isn't declared in the current scope
connect("my_signal", self.handler) # Nope, invalid get index 'handler'
connect("my_signal", self, "handler") # Nope, assigned type 'self' doesn't match the function argument's type (Callable)
connect("my_signal", Callable(self, "handler")) # Nope, "no constructor of 'Callable' matches the signature 'Callable(self, String)'

[ttencate]

I've already spent more time on this than I should have, and the rabbit hole is deeper than I expected.

I think having a confusing error message is much better than silently nullifying arguments, so in my opinion this PR is still an improvement over the status quo. But if everyone shared that opinion, the nullifying would never have been added in the first place.

I'll leave this PR open for discussion.

[vnen]

I think this is correct, we should revert the change. It should not silently convert the value to null. We need to find a way to improve the error message.

[vnen]

I just noticed that I was the one who merged this. I guess I didn't fully understand the consequences back then.

[akien-mga]

Thanks!