[curves] Curve25519 naming / renaming

[Trevor Perrin]

On a different list, Dan Bernstein suggested renaming the Curve25519
ECDH function [1] to "X25519", and taking the name "Curve25519" for
the underlying curve.

Since cryptographers are increasingly discussing the underlying curve
it makes sense to have a distinct name for it.  But I don't think we
should rename a function that's become so widely known to
implementors.

Cryptographers are a small tribe of smart people.  If the curve was
given a specific name ("P-25519" or whatever), they will quickly
adapt.

Implementors are numerous and more easily confused.  We've gotten used
to calling this specific function "Curve25519".

Dealing with an "X25519" function that is the same as the old
"Curve25519", but not necessarily compatible with uses of the new
"Curve25519", seems unnecessarily confusing.

So while clarifying the names here is a great idea, it seems better
not to change the name from its most common use.

Trevor


[1] http://cr.yp.to/ecdh/curve25519-20060209.pdf