it looks strange that we need to adapt float to integer for the comparison ...
I don't think that we can skip entities without tag because they have an impact on the clustering, for example outer-entities without tagged inner-entities will be clustered together
I am trying to produce some distribution maps and some of them are taking a while to compute.
Most of the time is spent in the clustering algos for the comparisons of floats
Maybe we could skip the clustering for entities without any tag?
I am not sure about this proposition but this is an idea I have