Multi-Agent Reinforcement Learning: Methods, Applications, Visionary Prospects, and Challenges