Describir: Bayesian Auxiliary Variable Models for Binary and Multinomial Regression