Solving Richly Constrained Reinforcement Learning through State Augmentation and Reward Penalties