Preconfirmations node does not handle catchup correctly

[jbearer]

If the preconfirmations node is reset or started late, it starts in a de facto catchup mode, where the L2 batches being synced via preconfirmations are well behind HotShot, and more importantly, well behind the batches verified on L1 by the regular node. When this is happening, we get many errors due to batches referenced by verified batches not being synced yet: https://app.datadoghq.com/logs?query=host%3A%22%2Ftestnet%2Fsepolia%2F%22%20service%3Acloudwatch%20%40aws.awslogs.logGroup%3A%22%2Ftestnet%2Fsepolia%2F%22%20%40aws.awslogs.logStream%3A%22zap%2Fpreconf_zkevm%2F173a1cca6d2f431c870910a29006abe0%22%20%22violates%22%20&cols=status%2C%40logger.name%2C%40error.message&context_event=AYr7TL2QAABRlCMVdimp5ABE&event=AgAAAYr-p_EEQ-LqxAAAAAAAAAAYAAAAAEFZci1wX2JEQUFETWVxT0lVekVUU0FBagAAACQAAAAAMDE4YWZlYTgtMjIxMC00ZmJjLWE5NjMtMzE1MzExYThiYzM1&index=%2A&messageDisplay=inline&refresh_mode=sliding&saved-view-id=2130045&storage=hot&stream_sort=time%2Cdesc&viz=&from_ts=1696403216408&to_ts=1696489616408&live=true.

These errors seem to cause problems with the catchup process itself. I believe a fix would be to not start syncing L1 blocks until we have synced L2 batches up to the HotShot head.

[jbearer]

A good test for this would be to just start the regular node, let it run for a while, start the preconf node, let it catch up for a while, then submit a transaction and wait for a receipt from the preconf node