Proper replacements

[ZanCorDX]

📝 Summary

• Parallel sealer killed (config parameter max_concurrent_seals gone)
• Removed every check for increasing TBV
• Removed every check for increasing competition bid
• Added a UnfinishedBlockBuildingSinkMuxer at the output of the BlockBuildingAlgorithms to choose the best block among the last block from each algo
• Added cancellations in relay submiton (/relay/v1/builder/blocks?cancellations=1)

💡 Motivation and Context

Good order replacement is critical!

✅ I have completed the following steps:

• Run make lint
• Run make test
• Added tests (if applicable)

[dvush]

@ZanCorDX I don't know if relay respect it but we need to submit with a url param to enable this https://github.com/flashbots/geth-builder-old-unused/blob/b6a6dd336988d3f866a6c611a6505885966ba189/builder/relay.go#L142

[github-actions]

Benchmark results for 4cf7956

Report: https://flashbots-rbuilder-ci-stats.s3.us-east-2.amazonaws.com/benchmark/4cf7956-feafa9e/report/index.html

Date (UTC)	2025-02-04T22:50:25+00:00
Commit	4cf7956fd36a6b87831fefaf79e3788f9f437f24
Base SHA	feafa9efd9988ee039d71cdfd8cb60de555d6f47

Significant changes

Benchmark	Mean	Status
gather_nodes_clone	-14.91%	Performance has improved.
MEV-Boost SubmitBlock serialization/JSON encoding	57.96%	Performance has degraded.

[ZanCorDX]

@ZanCorDX I don't know if relay respect it but we need to submit with a url param to enable this https://github.com/flashbots/geth-builder-old-unused/blob/b6a6dd336988d3f866a6c611a6505885966ba189/builder/relay.go#L142

The code will never cancel, it replaces. From what I read on the relay spec (https://flashbots.github.io/relay-specs/#/Builder/submitBlock) it should work:

Any new submission by a builder will overwrite a previous one by the same builder_pubkey, even if it is less profitable.

[dvush]

nit, to make code simpler and remove this option you can set it to B256::ZERO

[dvush]

also, would it make sense to have this as a feature flag? I am not against this but there may be a problem if it does not work for some reason and we would need to quickly update builder net binary instead of config

[ZanCorDX]

also, would it make sense to have this as a feature flag? I am not against this but there may be a problem if it does not work for some reason and we would need to quickly update builder net binary instead of config
Mmmm... that would take me a while. Maybe for the future?

[ZanCorDX]

Changed it to use a hash from tx hashs to avoid coping stuff.