Learning in Zero-Sum Team Markov Games Using Factored Value Functions