Data-driven approximation of distributionally robust chance constraints using Bayesian credible intervals