Major update, changes the way that the data-driven reference sets are determined, lets there be a reference set for each covariate instead of one reference set per analysis. This typically means the results will align closer with radEmu, although computation will take longer for more complex models with more covariates.