feat(protocol): Even more basic multi-prover with PoA

[Brechtpd]

Simplify oracle/system prover system, in a way that will be compatible with https://github.com/taikoxyz/taiko-mono/issues/13693:

• Blocks need a zk proof + a signature (currently just one we generate and share somehow)
• For proofs that need to be skipped only a valid signature is required, the zk proof verification is skipped. Here we force the proof time to be the target proof time to not impact the base fee calculation.

Then, currently optional because not needed, we just have an additional fail safe mechanism to overwrite and lock a fork choice directly, not going through the normal proving code in proveBlock.

[dantaik]

Maybe we should consider the SGX integration when we change the protocol interface. Brecht, do you know what will need to be verified onchain for SGX? Another question regarding SGX is how it impact our multi client strategy. Will the protocol be aware of the existence of multiple clients and has to handle each differently?

[Brechtpd]

Brecht, do you know what will need to be verified onchain for SGX?

Just an ECDSA signature, which could even sign the same hash that we use as the circuit input. So onchain the approach in this issue or using SGX will look exactly the same, just a signature that needs to be verified.

Will the protocol be aware of the existence of multiple clients and has to handle each differently?

They will all look the same on-chain, just a signature per node likely (to not increase the load per SGX having to run multiple nodes). We should probably even abstract everything away in verifier contracts, the protocol doesn't need to know if it's a zk-proof or something else. The protocol would just need to know it needs N proofs corresponding to some verifier contracts.

[adaki2004]

Simplify oracle/system prover system, in a way that will be compatible with #13693:

* Blocks need a zk proof + a signature (currently just one we generate and share somehow)

So block can be verified only if have both ? (Are we talking about testnet too ?).

* For proofs that need to be skipped only a valid signature is required, the zk proof verification is skipped. Here we force the proof time to be the target proof time to not impact the base fee calculation.

Ok so basically in such case set the proof time as proofTimeTarget. But are they overwriteable or not ? (I guess we run them not the community (?)).

Situations/scenarios we handling now, how would they look like in this setup ?

1. Can skip 9 real proofs and require 1 (for every 10 block) - but oracle proofs are not verifiable but have to be overwritten. 9 signatures coming from our node - and they verifiable afterwards or still needs (overwriteable) oracle proofs OR real proofs ?
2. System proof is verifiable (but having cool down time to be able to overwrite it): How we signal, this block is OK - dont need to overwrite it ? (Maybe with a different node's signature?)

[adaki2004]

Meeting minutes:

How to rearrange / map out this same functionality ? Oracle: throw away, System: throw away
When a prover wants to prove:
1. It queries the signature from the provider
2. Provides with normal signature (signature: same but graffiti and prover, but that to be double checked so just go with them first !).

Structure: array of proofs 1.: normal, 2: signature

Modify ForkChoice: allow some address to modify FK directly - and then lock it. (Only non-verified proofs). (Failsafe mechanism only).

[dong77]

My understanding is that we need 3 types of proofs (TypedProof):

1. ZKP
2. SignatureProof
3. SuperSignatureProof

A block is verifiable if it has (3) or the combination of (1, 2). Technically, 2, and 3 should be using the same code, but probably different signers. If this is true, I'm not sure what's the best way to express the such logics in our smart contract.

[Elpadronegeorge]

It would be easy to interact with less proof-tie time, the relayer should be able to interact directly .