Curio + Forest integration

[LesnyRumcajs]

Summary

It was identified that Forest should become a block producer. The gap is probably short to close (all RPC methods are already supported), so this is matter of exhaustive testing and resolving a couple of warts.

Motivation

Genuine client diversity.

Tasks

• Curio + Forest integration in a local devnet - worth testing under different scenarios, perhaps using Antithesis compose setup?
• Integrate the Curio + Forest in devnet in CI (some git necromancy might help)
• Workaround known discrepancies - the workaround here is intentional, as Forest might be right in doing certain operations the way it is doing, but it doesn't matter if the majority of nodes disagree with that. Some are obviously genuine fixes.
• Test on calibnet (somehow, to be decided how, the hardware requirements are pretty high)
• Test on mainnet (somehow, to be decided how, the hardware requirements are pretty high)
• Documentation with hardware requirements, some helpful scripts etc.
• Once at least one block is mined with Forest's help on calibnet/mainnet, publish a big fat post on SM for eternal glory and upvotes.

Risks & Dependencies

• Hardware requirements are pretty high on calibnet and mainnet. Maybe CS Infra could help. We cannot in good conscience recommend to users Forest as a block producer after testing exclusively on devnet.
• Curio should be fine now. Back when we started our first attempt it was under rapid development so it was difficult to catch that rabbit. Plus, we fixed a tons of RPC bugs along the way.

Additional Links & Resources