Gravar-mail: Concealing treatment allocation in randomised trials