[noise] Spec update - plans for Feb

[Trevor Perrin]

I'd like to roll some recent ideas into a spec update, seems like
these should be doable in February:


(1) Allow "strict" DH functions which return an error for invalid
public keys (secp256k1, P256, etc)

https://moderncrypto.org/mail-archive/noise/2017/000868.html


(2) Require strict DH functions to specify an explicit null value
which can be used for dummy public keys (same URL above).


(3) Specify "extra key derivation" from the key k.

https://moderncrypto.org/mail-archive/noise/2017/000869.html


(4) Modify pattern language for fallback patterns so the roles aren't reversed.

https://moderncrypto.org/mail-archive/noise/2017/000874.html


(5) Maybe move the security properties / identity hiding sections to
an appendix, since they're not that readable?


Anything else?  It would be great to have a NoiseSocket spec as well,
but that should be a separate doc, anyone want to try writing it?

I'm also hoping we'll see a better post-quantum lattice proposal soon,
which should unblock some of the work on hybrid forward secrecy.


Trevor