Independent Learning in Constrained Markov Potential Games