op-batcher: stay in sync with op-node rollup config

Upon startup, the op-batcher fetches the rollup config from the op-node over RPC. It never fetches the rollup config after that.

https://github.com/ethereum-optimism/optimism/blob/40554025301e5adde90191adba6ad432e185fa50/op-batcher/batcher/service.go#L238-L253

This is problematic during hard forks when the system deposit transaction format changes: an outdated rollup config will result in an error from which the batcher cannot recover without restarting.
https://github.com/ethereum-optimism/optimism/blob/40554025301e5adde90191adba6ad432e185fa50/op-batcher/batcher/driver.go#L310
```
failed to parse L1 info deposit tx from L2 block: data is unexpected length: 178
```

The op-batcher should stay in sync with the op-node's rollup config.

One solution is to fetch the rollup config if the above error is hit. An alternative is to for the batcher to restart itself automatically.

Read op-batcher/batcher/service.go around lines 238-253 to understand the startup rollup-config fetch, then inspect op-batcher/batcher/driver.go around line 310 and the reported deposit-transaction error. Define how the batcher should refresh or recover when the rollup config becomes outdated, and verify that it can continue after the hard-fork format change without a manual restart.

Written by the indexing model from the issue text.