State checkpointing for deployments

[epociask]

deployment.ts is prescribed as the primary script for performing rollup creator deployments onto mainnet environments. However this script doesn't appear to perform any state checkpoints or backups if failed during execution - meaning that a subsequent invocation would start from the beginning despite real txs having already been executed.

There are few cases that can cause this script to crash arbitrarily after executing some number of txs; e.g:

• parallel usage of the deployer account somewhere else that causes nonce misaccounting
• deployer account has insufficient funds after executing
• execution process crashes arbitrarily

It'd be beneficial to have some programmatic countermeasures in place to mitigate this risk; e.g:

• state checkpointing that dumps deployed artifacts up-to the failure point
• migrating the deployment script to using forge instead

cc @0x0aa0