Updates the way that the system config validates the resource
config to ensure that its not possible to set config values
that break the system. It is possible to set config values
that would burn too much gas, rendering the system unusable.
Every error cannot be caught, so try to catch the most possible.
Altering the default resource config is not recommended, but it
is possible.

This also adds fuzz tests that cover the problem, if the change
in the require statement in the system config contract is reverted,
then the fuzz tests catch the `UNDEFINED` error quite quickly.

op-node: finalize while syncing

ci: Reduce executor size

op-e2e: fix TestLargeL1Gaps flake

Revert "[op-node] Add support for non-batched RPC calls when batchSize == 1"

Our CCI usage more than doubled over the last few months. After looking at the config, we were using xlarge VMs for Docker build/publish steps as well as Slither which cost 40 credits/minute. I reduced those to values I think are more appopriate, and reduced some of our Docker instance sizes as well.

op-e2e: Longer seq window for TestSystemMockP2P

op-program: Fix l2 engine API to avoid fetching L2 blocks after the agreed head

op-batcher: Add Channel Bytes Counters

[op-node] Add support for non-batched RPC calls when batchSize == 1

[op-batcher] Bug: panics on nil access

fix(specs/proposals): Sherlock fixes

CI: Increase go lint timeout to 5m

op-program: Fix l2 engine API to avoid fetching L2 blocks after the chain head as part of importing.

op-program: Modify L2 oracle to not return error

op-e2e: Increase TestMockP2P timeout

op-program: Implement oracle based L1 fetcher