Simplify reifyTypeable

[treeowl]

In GHC 8.0 and later, KnownNat implies Typeable so we can implement reifyTypeable much more easily, and probably also considerably more efficiently.

[treeowl]

The reason we need the Typeable a constraint is now much clearer to me: s -> a means that s must have an a in it. But that means s will only be Typeable if a is as well. That's annoying. We could offer

reifyTypeable' :: a -> (forall (s :: Nat). (Typeable s, Reifies (Stable s a) a) => Proxy s -> r) -> r

That feels a bit weird. With a bit of annoyance in the compatibility department, I'm pretty sure we could clean it up:

reifyTypeable' :: a -> (forall (k :: Type) (s :: k). (Typeable s, Reifies (Stable s a) a) => Proxy s -> r) -> r

[ekmett]

This does have the downside of forcing a kind change on s in GHC 8+, so users would have to use polykinds or manually case the kind off of version numbers.

We could use a small type level shim to convert the Nat down to something of kind * and then that small wart would shave off.

[RyanGlScott]

Sorry for taking so long to look at this. (I think this PR predates me becoming a co-maintainer of the library.)

The new implementation of reifyTypeable looks great. I'm not quite sure what the implementation you had in mind for reifyTypeable' (alluded to in #45 (comment)) is, however. Can you post it here?

[treeowl]

I'll have to page this back in....