Quantum Multi-Agent Reinforcement Learning via Variational Quantum Circuit Design