Initial implementation of OpenPGP EdDSA and usage of the Curve25519 in ECDH

[filipnavara]

Also implements broader - but still incomplete - support for S expression keys in non-canonical format.

The endianness of Curve25519 keys in ECDH is a mess. Unlike all other ECDH keys the points are represented as little-endian for private keys. This was cross-checked with GnuPG. Some discussion about the standards is at openpgpjs/elliptic#1 (comment).

[michakinchen1988]

Deploy

[oseack]

Is there a plan to merge this? Thanks.
If you are busy and need help, please let me know.

[filipnavara]

I haven't heard a single word from the maintainers. Until that happens we are running a fork for ourselves.

cc @peterdettman

[oseack]

@filipnavara I really appreciate your work. Hope this can get merged some day.
@michakinchen1988 Any plan to make it happen? Looking forward to this fix. Thanks

[peterdettman]

@filipnavara Was this based on the equivalent code in bc-java or written independently?

[filipnavara]

Independently, bc-java didn't have the functionality when I wrote it (not sure if it has it now).

[oseack]

@peterdettman Do you prefer bc-java equivalent?

[peterdettman]

@oseack We do prefer to keep the codebases similar for ease of maintenance, but I am not asking for anyone to redo this. I will just substitute any bc-java stuff where appropriate.

[oseack]

@peterdettman that'd be great. Please let me know if you need help!

[fnajera-rac-de]

Thanks @filipnavara for your code! I have put your changes on top of the current master and also migrated the corresponding Java tests - which are also successful (see fnajera-rac-de@f10df3a).

[peterdettman]

As mentioned at #345, the EdDSA/XDH stuff has been ported from bc-java.

Comparing to this PR, the only thing missing appears to be the SExpr class changes, which I plan to come back to after we get the v2 release out.