Could it be that you missed the step

If P receives an Election message from another process with a lower ID it sends an Answer message back and starts the election process at the beginning, by sending an Election message to higher-numbered processes.

in your implementation? I was wondering why at times when process 0 was going through the election and then elected a leader, and at the same time I was starting up process 1, process 1 might get a victory message from 0 and thus declare the leader is 0. And then both would go on pinging each other and merrily keep that state.

Your MessageListener doesn't have the capability to determine the process id of the sender of an election, and there's nothing in there retriggering an election should an election message be received from a peer with a lower Id.

Also, I was wondering, what's the reason for the 2 second delay between starting the MessageListener and the LeaderElection - is that a safety buffer to work around the problem I mentioned above ?

I went ahead and implemented that missing step. Plus I was occasionally getting the wrong leader when the process with the lower id was in the middle of the election when the process with the higher id started. So as a safety, when I get an election from a lower process, I retrigger the election which corrects the state.