Mastering Chess and Shogi by Self-Play with General Reinforcement Learning