Discovering General Reinforcement Learning Algorithms with Adversarial Environment Design