Please excuse me if my suggestion makes little sense.
I was looking into constrain issue and came across this How to include "constraints" in the optimization process [Solved] - Grasshopper
One of the suggestion is use Null output. When I was experimenting with this, it seems that wallacie does not remember that x set of genes group resulted in null and tries again in future searches. It would be helpful if in future generation we can set it to not try same set of gene value for a given group.
Thank you for wonderful work.