Many raft timeouts during scale testing

When doing scaletesting on ec2 machines with a volume backed root-disk, the runOnLeader method hit a lot of timeouts:

 case <-s.config.Clock.After(s.config.ForwardTimeout):
  logger.Infof("timeout")
  s.record(command.Operation, "timeout", start)
  return lease.ErrTimeout

Perhaps we should consider some form of batching to send the requests to the raft leader.