op-node: Give `admin_startSequencer` a chance to sync to requested head

The sequencer admin rpc `admin_startSequencer` currently directly calls `Sequencer.Start`. But if the candidate sequencer is still processing the requested head, this will fail. This could be improved by giving the sequencer a chance to sync up to the head first, within a short timeout window.

Start at the op-node admin_startSequencer RPC and trace its direct call to Sequencer.Start. Determine how the candidate sequencer reports that it is still processing the requested head and where a short sync timeout can be applied. Done means the RPC gives synchronization a chance before starting and still fails cleanly when the head is not reached in time.

Written by the indexing model from the issue text.