Adapt ::aead to name change of NewAead to KeyInit

[nc7s]

I'm interested in using schnorrkel derived keys as DH keys, but the ::aead module seems unfinished: can not compile, lacks documentation, etc.

A first commit adapts to the name change of NewAead to KeyInit in aead crate (#90), while I look forward to finishing the module along the line.

[burdges]

This is just an update of the aead module? okay cool.

We should ideally discuss if this module is in a good enough form, especially if folks want to use it with key derivation.

We should provide a noise feature & module which implements noise_protocol::DH trait probably, but..

Almost anyone using this wants one-off messages not handshakes I think, so that's not noise. I suppose derivation provides an interesting option here, but not sure soft derivations play nicely.

[nc7s]

My intended use case would be that, with only a signing key of Alice, Bob can derive an encryption from it, while Alice can later derive the corresponding decryption key. Basically, to avoid a "standing" encryption key. There's no prolonged data exchange here, so it's indeed "one-off".

On a second thought, a PR isn't the best place to track something like this, and I'm no cryptographer either. Like, I don't really understand how Noise works, although it seems good to have.

[burdges]

Is this useful to you in its current state? I can try to figure out the failed CI, which may not be anything to do with you, and then go ahead and merge.. and other stuff can be another topic.

[nc7s]

Just saw the failures so tried to fix them. Also this is just a normal PR to fix a few things, no need to change your course ;)

[burdges]

I fixed the CI and merged this. I'm not sure what AEAD functionality would be possible without getrandom, but maybe something.