Text this: A Survey of Multi-Agent Deep Reinforcement Learning with Graph Neural Network-Based Communication