Mastering the Game of Stratego with Model-Free Multiagent Reinforcement Learning