Is it recommended to check for the presence of autocorrelation in the CI Bayesian Structural Time Series through the DW statistic, for example?