Optimizing Mastery Learning by Fast-Forwarding Over-Practice Steps