Multi-Agent Reinforcement Learning for Energy Networks: Computational Challenges, Progress and Open Problems