Structure in Reinforcement Learning: A Survey and Open Problems