Gravar-mail: Statistical considerations for pediatric multidrug–resistant tuberculosis efficacy trials