Stop synchronizer blocking the sequencer

[arnaubennassar]

There is a window of time in which the sync blocks the sequencer:

• From: a sequence is mined on L1, and the sequencer realises that the last virtual batch in the SC is greater than the one in the DB
• To: The synchronizer catches up with the SC adds the latest virtual batch on the DB (after executing them!)

In a setup where there is sustained load on the network, this means that basically we've to process all the txs twice, making the sequencer work in intervals.

The sequencer should only wait if the last batch num < last virtual batch. If the trusted state is ahead of the virtual state there is no reason to wait, if something goes wrong, there will be a L2 reorg.