Apply KEM from RFC 9180; TODO authenticated KEM

[sander]

I have not fully checked the details yet, but suspect that the ARKG spec can be simplified by referring to the KEM from RFC 9180.

It removes the ability to provide a context string, but I don’t think that is a problem, since the output from KEM-Encap is ephemeral and therefore bound to the application context anyway.

Referring to DHKEM specifically does add "HPKE-v1" to the HKDF input keying material, which is awkward. But there seem to be already precedents of Internet-Drafts and RFCs reusing the DHKEM outside of the context of HPKE.

Probably the section “Using ECDH as the KEM” can be removed as well, if we find a predefined KEM for secp256k1.

[emlun]

Sorry for leaving this unaddressed...

I think this seems like a good idea, but like #27 it's unfortunate that it's a backwards-incompatible change that affects implementations. We may be too far along with concrete prototypes at this point, I'll have to ask our firmware team what they think of it. But if we're going to do this, we should do so before we define official algorithm identifiers (as opposed to the private range COSE algorithm identifiers our prototypes are using for now).