Rafael Harth comments on Conditions for Mesa-Optimization