Implement Schnorr signature encoding according to SEC1 v2 (v2.3)

[Geod24]

See section v2.3 of this paper: https://www.secg.org/sec1-v2.pdf
This is not urgent, but required before we release publicly, as we're dependent on Libsodium's internal state at the moment.

Related to #208

[hewison-chris]

I spent some time researching the need to use a different encoding (SEC1 v2) for the signature as generated by the libsodium library. From what I can see the checks in the verify code are to prevent the malleability of the signature.

[hewison-chris]

In my opinion we should continue to use the libsodium 64 byte encoding of the Signature (R, s).

[Geod24]

Agreed. According to what @hewison-chris found, the points are already stored in canonical form.