I could be wrong, but I believe a Schulze method would solve that. The compromises could remain independent pieces which pass or don't pass depending on a prefe

I could be wrong, but I believe a Schulze method would solve that. The compromises could remain independent pieces which pass or don't pass depending on a preference matrix.