Cluster enlarge period

[ugurdogrusoz]

Currently, we seem to check whether or not we should enlarge a cluster in every iteration. This is not very reliable since the situation might change over iterations similar to other heuristics like swapping neighboring nodes on a circle. So we should accumulate such data over a number of iterations and enlarge it if needed.
We should also make sure that this check is not done redundantly as it's done currently for every in-circle node but only once per cluster! I think we should move all the nodes first and then do this enlarge check with a certain frequency.