Simultaneously Updating All Persistence Values in Reinforcement Learning