Our code (library, KDC) should be parsing and processing at least the authorization data container types, even if we don't understand any real authz types yet. We should also provide library helper routines for constructing or breaking apart these container types' encoded forms.