Improve logging of topology updates

Rebalance confirmations are logged with the event logger, but topology updates aren't. It should be the other way around, because the number of confirmations grows with the cluster size, and the number of topology updates only depends on the number of caches and joiners/leavers.