Analysis of Multiscale Reinforcement Q-Learning Algorithms for Mean Field Control Games