Optimal Design Of Cluster Randomized Trials With Binary Outcomes