Change validator key to BLS12-381

[brunoffranca]

What ❔

Changes the validator keys to use the BLS12-381 curve.

Why ❔

Now that we decided to separate validator and attester keys, we can use a more efficient curve for the validator keys.

[dnkolegov]

@brunoffranca, does that mean that the bn254 curve will not be used at all?

[brunoffranca]

It will be used to sign L1 batches. L2 blocks and all consensus messages will use BLS12-381.
After EIP-2537 gets merged, everything will use BLS12-381.

[brunoffranca]

@pompon0 I added two more commits removing MsgHash. Signature schemes already hash the message internally, it doesn't make sense to be hashing twice. See if you approve.

[pompon0]

@brunoffranca the point of having MsgHash was to avoid passing the whole message from place to place, and to avoid cloning as the messages can get quite large. I agree that so far we have underutilized those optimizations, but I can already see that you had to add a couple of extra clones in this PR already.

[brunoffranca]

the point of having MsgHash was to avoid passing the whole message from place to place, and to avoid cloning

@pompon0 Do you believe that this extra cloning does make a performance difference? And that we will actually start utilizing this MsgHash construct more in the future? If yes, I can remove these two commits.

[pompon0]

Do you believe that this extra cloning does make a performance difference? And that we will actually start utilizing this MsgHash construct more in the future? If yes, I can remove these two commits.

It is hard for me to tell. We can reintroduce it later, once there is a need to do some serious profiling. Just please avoid introducing Clone requirements in generic functions.

[pompon0]

added some comments

[pompon0]

also FYI, if we reintroduce it later it will be a breaking change to the protocol.

[brunoffranca]

Ok, like I always say, tie break goes to the reviewer. So I'll remove these two commits.