make wasm32 compatible

[DougAnderson444]

While trying to use multikey in WebAssembly code, I realized sodiumoxide doesn't compile to that target, and should be upgraded to the RustCrypto crates as it uses the older 8 bytes nonce instead of the latest 12 byte nonce.

I took some liberty to make some small clippy clean ups too.

I changed the serde tests, but didn't look too deeply into the details as they seem a bit fragile with the hard-coded checks in there. Recommend those serde tests be made into something a bit more robust eventually, like roundtrip or somehow codify the encoding to automate the checking. Anyway, the tests pass but should be looked at further I think.

In the meanime, this fixes issue #1 and compiles to wasm32 =)

[DougAnderson444]

I also ran into the issue of using ssh-key and ssh-encoding.

This cannot compile to wasm32* either, so I placed it behind a feature flag.

I also switch new_from_random_bytes to using dalek directly to remove it's dependency on ssh-key which is not needed.

[DougAnderson444]

Last blocking issue should be cryptidtech/multisig#1

Once that is merged, that dependency can be updated and this crate will compile to wasm32

[DougAnderson444]

I have excluded default-features on ssh-key & encoding, as that way we keep ssh but can build to wasm targets

[DougAnderson444]

When cryptidtech/multisig#1 get merged, I'll update this PR and make it no longer Draft

[dhuseby]

Your multisig PR was just landed. Flip this one out of Draft so we can move forward.

[DougAnderson444]

I updated to test test_serde_encrypted_secret_key_compact to reflect the new 12 byte chacha2-poly1305 nonce and 32 byte key.

I added various target checks in the CI, including wasm32

All tests passing now, should be good to go!

[DougAnderson444]

Sorry, fighting with Git a bit this morning. Github was showing conflicts, which I tried to resolve and now there's no conflicts showing. Weird.

Recommend running the CI again to confirm it's still good to merge and there are no conflicts.

[dhuseby]

It looks like there was a merge/clippy --fix error in all of these changes. Duplicated code everywhere. BTW, I like fn from(mk: Multikey) better than fn from(val: Multikey). This may be my fault because I was pushing clippy fixes this week but it's in your PR so you get to fix it 😂

[dhuseby]

Fix the SshError derivation to use thiserror and a lot of the changes go away because the error elision is possible at compile time. There's also a few changes you made that are incorrect. I think using standard macros like vec! and matches! produces faster code IIRC.

[DougAnderson444]

@dhuseby I believe I have addressed and changed all your points, can you have another look with the latest commits?

[dhuseby]

LGTM

[dhuseby]

🚀