Existence of Stationary Optimal Policies for Some Markov Renewal Programs