Cache document type spec information

document type specs are used both when reading the document type spec itself and when reading from and writing to documents. We expect that loading them involves extensive use of the repository's content type service, augmented with JCR read activity, and it seems to make sense to cache document type specs, since document types themselves are rather static (we discourage changing them in production).

Caching document type specs also means that changes to the underlying document type definition (in JCR) should flush the cache.