Crypto Pair trait refactory

[davxy]

While preparing the integration of BLS curves I noticed some duplicated code that can be directly part of the Pair trait default implementation.

Note 1. I've left a temporary test just to check that the result was the one expected for an implementation that was essentially doing the same stuff but in a fancier way (I'll remove it before merge)

Note 2. I would also like to take the opportunity to hear your opinion about substrate-bip39 repo/crate.
Is a library containing only 2 functions (both of few lines), and one of these now ends up being unused.
Maybe we can just embed the seed_from_entropy function in the core code (keeping the tests obviously) Any drawback?

[melekes]

👍

[bkchr]

Maybe we should add some debug assert that big_seed.len() > seed_slice.len()? Otherwise there may will be some implementation that will just use some zeros?

[davxy]

Yeah probably would be better to fail than filling the tail of the seed with zeroes.

But what about embedding the only function that substrate-bip39 offers directly into substrate?
As I said that library provides 1 function (the other is not used anymore after this pr).

The function is this:

pub fn seed_from_entropy(entropy: &[u8], password: &str) -> Result<[u8; 64], Error> {
    if entropy.len() < 16 || entropy.len() > 32 || entropy.len() % 4 != 0 {
        return Err(Error::InvalidEntropy);
    }
    let mut salt = String::with_capacity(8 + password.len());
    salt.push_str("mnemonic");
    salt.push_str(password);
    let mut seed = [0u8; 64];
    pbkdf2::<Hmac<Sha512>>(entropy, salt.as_bytes(), 2048, &mut seed);
    salt.zeroize();
    Ok(seed)
}

Sounds a bit bombastic to define a library only to provide a function...

Furthermore, we can then easily modify it to fill a mut slice (passed as parameter) to fill it with an arbitrary number of bytes... Thus there would be no need to check if we can fill the seed buffer because we always can

[bkchr]

I have no real opinion here. We can copy the function, I don't really care. I don't know the history behind the substrate-bip39 crate.