Refactor (typing): add annotations in dialect/parser/schema modules

[OutSquareCapital]

This typing PR annotate numerous functions/methods from core modules with more precise types.

Going from x: list to x: list[Any] or even x: list[object] may seem pointless, but strict type checkers like basedpyright with ALL rules, or pylance with strict mode=on , will complain on non-defined generic types, both at the definition AND call sites (hence for end users of the library).

For example:

[georgesittas]

@OutSquareCapital I think wherever we expect args we want to also handle None. There are some places (example) currently where this is not the case. I wonder if we should just do Sequence[t.Any] or Sequence[object] for now to avoid dealing with typing args precisely. What do you think?

[OutSquareCapital]

If we have a known set of types expected to be used in the Sequence for args, I could try to figure out a solution avoiding Any.
Otherwise yes, Sequence[Any] would be the solution.
(object can't be used here unfortunately since there's "concrete" attribute/method uses at some point)

[georgesittas]

I think it's ok to go for Any now if that's a strict improvement over what we have today. Analyzing the possible values for args and/or possibly restricting them as part of a refactor to get a cleaner typing API can be done as a follow-up to this.

[OutSquareCapital]

I think that's a good idea.
So, I refactored all the args: Sequence[...] to args: Sequence[Any], with a centralized type alias to clearly mark them, and make future changes easier.