feat: classes for inference systems and logical equivalence

[fmontesi]

This PR is the beginning of work for formalising general concepts about logic in CSLib, in order to streamline the development of logics. A major challenge is that judgements, proof systems, and semantics can be very different across logics, so we cannot assume any particular shapes for them.
To address this, this PR introduces:

• A class for inference systems, which gives access to the ⇓ notation for derivations and the associated notion of Derivability.
• A class of heterogeneous contexts (HasHContext), which allows for defining contexts that, when filled with a value of type T, yield something of a different type than T. This is useful for formalising judgemental contexts: a context with a hole that can be filled with a proposition, resulting in a judgement.
• A class for LogicalEquivalences. We formalise a logical equivalence as a congruence on propositions that preserves validity under any judgemental context.
• Instances of these concepts for HML and CLL, which serve as examples of very different structures.

Depends on #393 for the new context fill notation.

[chenson2018]

Just some nitpicks, I haven't properly thought about the big picture here yet.

[chenson2018]

We should be consistent within a single declaration of Type* versus explicit universes. I'd write

Suggested change

	class HasHContext (α : Type u) (β : Type v) where
	/-- The type of contexts. -/
	Context : Sort*
	Context : Type*
	class HasHContext (α β : Type*) where
	/-- The type of contexts. -/
	Context : Type*

[chenson2018]

I get the impression that part of fixing the performance issues in LTS can start with being more strict about Prop-valued def used with grind. I don't quite understand all the new changes around defeq and reducibility, but it seems in general things like this should be a structure or irreducible.

[chenson2018]

It's okay to write

Suggested change

	induction ctx
	<;> simp only [Proposition.Context.fill, Proposition.denotation]
	<;> grind
	induction ctx <;> simp [Proposition.Context.fill, Proposition.denotation] <;> grind

because grind is considered a tactic permissible to follow the flexible simp. (Though it's not clear to me what is missing that grind won't close this on its own)

[chenson2018]

As a reminder, you can use where with def as well.