Inverse Factorized Q-Learning for Cooperative Multi-agent Imitation Learning