Gravar-mail: Two-stage designs optimal under the alternative hypothesis for phase II cancer clinical trials