Max number of bad cluster iterations is too high

[justinian336]

It often happens that the number of bad clusters does not decrease no matter how much you keep iterating (the bad cluster fix does not change anything). In that case, waiting for 100 iterations before breaking out of the loop is too much. We should be able to exit this loop more smartly. Also, the algorithm should be robust to zero-nodes clusters, in the sense that K should be interpreted as a maximum number of clusters instead of the exact number of communities.

lighthergm/lighthergm/R/EM_wrapper.R

Line 502 in d426855

if (counter_bad_cluster >= 100) {