Improve Population CI for Hazard Ratio

Currently an issue is that the CI for the population HR and survival distribution is calculated by taking an average across the predicted value for all subjects within the population. This is unfortunately does not produce accurate confidence intervals for the population.  A paper with proposals to address this can be found here: https://onlinelibrary.wiley.com/doi/full/10.1002/sim.8713