Remove Kyber90s and Dilithium-AES variants

[dstebila]

NIST has said they have no intention to standardize these variants. I checked with Peter Schwabe and he's not aware of any plans to keep them alive elsewhere.

[dstebila]

As suggested by @baentsch, easiest if we save doing this until we are also updating other algorithms, so as to only have to do one downstream update.

[baentsch]

As suggested by @baentsch, easiest if we save doing this until we are also updating other algorithms, so as to only have to do one downstream update.

Allow me to reconsider my suggestion as also the IETF interop test team by now has suggested to drop these variants ASAP. Further considering that our release cycles are pretty long, I wouldn't mind implementing this right now (also in the downstreams) to have it done by the time #1420 finally gets resolved.

[dstebila]

Sure, fine with me.