Is it necessary to eliminate the posterior dose margin in prostate brachytherapy to achieve an acceptably low risk of late rectal morbidity?