Fail in processSequenceBatches if we don't have the parent batch

This will cause an error if there has been a reorg since we fetched the
batches. In normal mode, this should never happen, since we check for
reorgs and then fetch batches synchronously. But in preconfirmations
mode, the batches are fetched asynchronously wrt to the reorg
detection, and so the L2 head can change under us. In this case the
new check will fail and we will re-fetch preconfirmations after
checking the new L2 head.

Fixes EspressoSystems/espresso-polygon-zkevm-demo#246

synchronizer/synchronizer.go

@@ -611,6 +611,23 @@ func (s *ClientSynchronizer) processSequenceBatches(sequencedBatches []etherman.
  return nil
  }
 
+ if lastBatch, err := s.state.GetLastBatchNumber(s.ctx, dbTx); err != nil {
+ log.Errorf("Error fetching previous batch number: %w", err)
+ rollbackErr := dbTx.Rollback(s.ctx)
+ if rollbackErr != nil {
+ log.Fatalf("error rolling back state. rollbackErr: %s, error : %w", rollbackErr.Error(), err)
+ }
+ return err
+ } else if sequencedBatches[0].BatchNumber != lastBatch + 1 {
+ err = fmt.Errorf("New batch %d is not successor of last synced batch %d, there may have been a reorg", sequencedBatches[0].BatchNumber, lastBatch)
+ log.Warnf("Possible reorg detected: %w", err)
+ rollbackErr := dbTx.Rollback(s.ctx)
+ if rollbackErr != nil {
+ log.Fatalf("error rolling back state. rollbackErr: %s, error : %w", rollbackErr.Error(), err)
+ }
+ return err
+ }
+
  for _, sbatch := range sequencedBatches {
  // Ensure the L1 origin for this batch is in the database. Since the L1 origin assigned by
  // HotShot is not necessarily the same as an L1 block which we added to the database as a