Finish up support for Newtypes

[robdockins]

The cryptol remote API still needs some work to be able to handle some unique issues that arise from newtypes (basically, the generative nature of newtype declarations). See:

cryptol/cryptol-remote-api/src/CryptolServer/Data/Expression.hs

Line 343 in f55aea6

TVNewtype _u _ts _tfs -> liftIO $ throwIO (invalidType (tValTy ty))

In addition, I think we should consider if newtypes should be allowed as property arguments. I'm starting to be inclined to say they should not be allowed, as this might break internal invariants. However, it makes them considerably less convenient, without something like #1028 to make it easier to program generation code for newtype values.

[robdockins]

CF #1037.

This ticket should really just be for finishing the remote-api work.

[pnwamk]

Initial support for newtypes and other complex values has been added to the cryptol server: #1200