"fingerprint" and "id" inconsistent terminology and possibly inconsistent specifications

Apparently "identity" might mean a SHA1 of a DER-encoded RSA key or an ED public key, possibly a KP_relayid_* or a KP_dirauth_id_* or something, and sometimes the term "fingerprint" is used.

And the representations (encoding) in netdocs are described ad-hoc throughout the spec, piecemeal.

Instead:

1. Keys that don't have names should have them.
2. We should use a formal notation to specify precisely what each netdoc field is. For keys (or hashes thereeof) we should use the proper key names. We should use notation (or perhaps brief terminology) to specify whether the netdoc protocol element is a hash, and if so what hash function.
3. We should avoid recapitulating encoding syntax; it should be defined somewhere central.
4. This means that the same approach should be taken everywhere throughout the netdocs. If it isn't (which seems likely) then this is a protocol deficiency but fixing it would be an incompatible change (depending on which elements).
5. We should avoid using the words "fingerprint" and "identity" unless we can give them definite meanings.

Prompted by !305 (comment 3122868)