Ok, new versions with rough MOI approximation. I'm sampling only the vertices, assuming a sampling 5000 (!) times the number of vertices. It's generating something that looks like a MOI matrix, but the numbers are stil way too high. Can anyone check it ingame to see how a model behaves?
I don't think I'll be adding a proper volumetric sampler, unless someone points me to very "copiable" code of a method that checks if a point is inside a model (which may not be perfectly closed, which complicates matters a lot). If the numbers are just too high, I can just divide it by something. If they're generally off, we have a bigger problem.
Anyway, the RAR with the new versions is attached.