Rollback command line argument

[eyusufatik]

We’ve had multiple times, on our own infra, that full nodes or the prover database got corrupted, we had to resync to the chain.

This won’t be acceptable on mainnet, what are we gonna do if the sequencer’s database gets corrupted?

Ideally, there should be two different rollback features.

1. Invoked with a cli arg like so

   —start-L2-sync-from L2_HEIGHT —start-L1-sync from L1_HEIGHT

   This will make sure everything above certain versions for the 3 databases we have will be deleted.

2. On startup, we check the latest versions of the 3 databases, and if we detect an anomaly, we prune the “higher” table, and start syncing form “lower” table.