Option to compress schema on encode

[encody]

Currently, schemas are encoded with type names in essentially plaintext. This is unnecessary; they can simply be identified by unique bytestrings—the human-readable name is superfluous and wasteful.