Handle enumeration of infinite types properly

[byorgey]

Since Disco is now strict, calling enumerate on an infinite type simply hangs as it tries to strictly construct an infinite list.

We should either (1) disallow enumerate on infinite types, or potentially (2) change enumerate so it returns a function N -> a (and make list do what the old enumerate did?).

[byorgey]

However, note that it is still useful to be able to enumerate infinite types internally, to be able to test propositions over infinite (but searchable) types. There's just no point to allowing the user to say, e.g. enumerate(N).

[byorgey]

When typechecking enumerate we simply need to make sure that the type is finite in addition to checking that it is valid.

disco/src/Disco/Typecheck.hs

Lines 1117 to 1119 in 82c9994

	typecheck Infer (TTyOp Enumerate t) = do
	checkTypeValid t
	return $ ATTyOp (TyList t) Enumerate t

Note that #175 is related, and it may make sense to fix them both at the same time.

[byorgey]

Alternatively, maybe we want to allow returning a function N -> a from enumerate, i.e. give direct access to what the enumeration library is giving us. In other words, enumerate should return a functional sequence rather than a finite list? See also #347 . Maybe list could be overloaded to do what enumerate currently does for finite types.