Autonomous Option Invention for Continual Hierarchical Reinforcement Learning and Planning