Empty tuple decontruction is wrongly parsed as correct

[Xanewok]

This happens because we use a Separated grammar node parsing an Optional item. Since a missing item is treated as a succesful parse, this doesn't fail.

We should make sure to support parsing (, ...) with at least a single item and/or comma.

For more context:

• () = () is parsed by Slang as correct but solc rejects it - tuple deconstruction can't be empty
• (,) = () is rejected by solc at parse level; at least one item should not be skipped/"empty"
• (a,,),(,a,),(,,a) are all valid deconstructions

[OmarTawfik]

Discussed f2f: let's keep it as-is, and leave such validation to type checking.