A Bayesian Difference-in-Difference Framework for the Impact of Primary Care Redesign on Diabetes Outcomes