Lucius Bushnaq comments on Computational Superposition in a Toy Model of the U-AND Problem