bug: sequencers forking

[d-roak]

There is currently a problem where the sequencers might fork from each other.
This issue needs more investigation because it's unclear how to reproduce it.

Some things that might cause it:

• NotInFinalizedChain: trying to revert the same finalized block hash paritytech/substrate#14347
• Slow block import paritytech/polkadot-sdk#13
• using the --force-authoring flag
• nodes storing different chains that were not finalized yet, and using that chain to sync instead of using the longest chain. e.g:
  • node1 has finality 500, highest 1000 and some block 600 stored in the db
  • node2 joins and syncs with the 600 instead of the 1000

Edit (26/06/2023): downgrading substrate from 2e7fde8 to polkadot-v0.9.43 fixes the issue

[EvolveArt]

Downgrading substrate from 2e7fde8 to polkadot-v0.9.43 fixes the issue