Support BAL Devnet 3

[jsign]

This PR has changes to support bal-devnet-3 branches for ELs.

High-level overview of changes:

• Switch Ethrex to bal-devnet-3
• Switch Reth stateless crate to bal-devnet-3, which also means Reth-crates too.
• Add NewPayloadRequestAmsterdam to support Amsterdam accordingly (i.e. new ExecutionPayload types, etc). Also, correct deserialization and validations for both ELs, etc.
• I changed the blocks.tar.gz for CI with some blocks I pulled from the real BAL devnet 3 chain that went live just some days ago. Had to do a bal-devnet-3 for workload

Main goal here is:

• Have a branch to potentially prove bal devnet 3 to see BAL influence.
• Run EEST benchmarks filled for Amsterdam
• Have some things setups for potential sub-block proving experiments, which rely on BALs.

I think for a bit, this PR will be kept as a draft. At some point, we can think it makes sense to merge, but that can make things quite unstable, so I'm not totally sure about that. We could use features to gate things, but it can get hairy -- I think we can discuss this eventually if it ends up being the main branch we use soon.

[jsign]

@han0110, not ready for review or similar (although feel free to skim if you're curious).

Just leaving two particular comments here just fyi (nothing to do).

[han0110]

I think for a bit, this PR will be kept as a draft. At some point, we can think it makes sense to merge, but that can make things quite unstable, so I'm not totally sure about that. We could use features to gate things, but it can get hairy -- I think we can discuss this eventually if it ends up being the main branch we use soon.

Perhaps we can do the release like execution-spec-tests to release the ELF with BAL feature in a dedicated branch? So the main release can still be used for zkboost, and other branches can be used to do benchmark.