If you activated the covariates in the HB estimation dialog (advanced dialog), then the covariates were used to estimate the utilities that were written out to the .hbu file. Respondents were smoothed to their cohort's population means, according to the covariates. Thus, the utilities written to the .hbu file were influenced by the covariates.
When you run market simulations in SMRT, it just reads in the part-worth utilities (the point estimates found in the .hbu file) and uses those to predict for each respondent the probability of choice for alternatives in a market scenario.
Under this method of market simulations using the posterior individual-level part-worth utilities (the lower level betas), there is no need or reason to activate the covariates as predictive weights within the market simulator.
You can compare the predicted shares of preference (shares of choice) from the population to your held out shares of preference for hold-out validation.
As an aside... if you wish to use the covariates as segmentation, filters, or respondent weights within the SMRT simulator, you may import those variables as segments or weights. Within SMRT, click File + Merge Variables and follow the prompts to merge segmentation or weighting variables into your market simulator project. After doing that, then within a market simulation scenario, you can use the "Banner" drop-down (top left corner of Market Simulator dialog) to cut the results by values of a categorical segmentation variable. Or, you can apply a merged variable as a respondent weight by editing a market scenario and clicking "Respondent Weights" (unchecking "Equal " and defining a segmentation variable as a weighting variable using the "Weights" button.