Gravar-mail: Sample size considerations in the design of cluster randomized trials of combination HIV prevention