feat: generalize constructor generation further

[Equilibris]

We need a more general notion of constructor generation. The features we need are:

• The ability to change the output type to the desired type
• The ability to control what a recursive type should be able to be replaces with (for DeepThunk generation)
• The ability to add required binders used in for-example explicit types (in the case of DeepThunk this is the recall parameter)

In this PR we add these features while keeping the behaviour for the specialized case unchanged. This PR is non-functional in place of current code

[AtticusKuhn]

Looks good to me.

[AtticusKuhn]

I would prefer a more descriptive name than "x".

[alexkeizer]

This PR is non-functional in place of current code

I guess what you mean here is that you generalize a bunch of things, but don't use the extra general capabilities, leaving the existing behaviour unchanged? I would reserve the claim "non-functional change" for PRs that really only move around definitions, or change the way somethings is phrased to an alternative, but 100% equivalent statement.

Here, I would instead flag it as something akin to: "we don't yet use the more general capabilities (but require them for a future PR), so the behaviour of existing code is unchanged".

[alexkeizer]

I agree with Atticus' comment about having a more descriptive name, and would like you to phrase the docstring a bit more clearly.

Otherwise, LGTM

[alexkeizer]

Suggested change

	The binders param allows for adding constant binders to an expression.
	Any binders passed in `binders` are added as parameters to the generated constructor

[alexkeizer]

Suggested change

	(binders : TSyntaxArray ``Parser.Term.bracketedBinder)
	(binders : TSyntaxArray ``Parser.Term.bracketedBinder := #[])

This seems like a good place to use a default value (but feel free to ignore if you disagree)