Gravar-mail: Bayesian analysis of heterogeneous treatment effects for patient-centered outcomes research