Dynamic Leader-Follower Consensus with Adversaries: A Multi-Hop Relay Approach