Introduce PAYLOAD_BUILDER_VERSION

[jtraglia]

This PR introduces a new constant (PAYLOAD_BUILDER_VERSION) for execution payload builders. Now that EIP-8282 is included, builders are not restricted to version 3 so this can start at 0. Builders onboarded at the fork are now hardcoded to use PAYLOAD_BUILDER_VERSION instead of the withdrawal prefix. And checks are added in gossip & process_execution_payload_bid to ensure the version is correct.

[potuz]

This I think it's more or less the same comment I had on the other PR. If we do allow builders to be created with any version, then off protocol projects may want to use that. What's the reason to restrict propagation to only 0x00 in this case? I would prefer to only restrict one reserved space and enforce this at creation

[jtraglia]

If we do allow builders to be created with any version, then off protocol projects may want to use that.

If there's a will, there's a way. There's nothing we can do to prevent this. For instance, instead of the version they could use the first 4 values of the execution address. Ie if builder.execution_address[0:4] == "0xabcd".

What's the reason to restrict propagation to only 0x00 in this case?

Because this check also exists in process_execution_payload_bid and there's no need to forward a bid which would be considered invalid.

[potuz]

Because this check also exists in process_execution_payload_bid and there's no need to forward a bid which would be considered invalid.

Yes, I think the assert on process_execution_payload_bid should be removed if we accept any version on deposits. We can get right what we didn't with validators now.

Allowing validators with different withdrawal prefix was useful (we are using it now to onboard builders for example).

Reserving a space for builders versions is also useful (we can do whatever we want with those reserved ones).

[jtraglia]

I'm going to make an executive decision & merge this PR.