Refactor: Compute epistemic uncertainty for quantiles

[AuguB]

In addition to providing the centiles, the evaluate function should also output the eptistemic uncertainty about those centiles. In the case of HBR, it is clear that the variance along the chain can be taken as the epistemic uncertainty. We need to come up with a universal measure that can be implemented for HBR, BLR, and GPR.

[AuguB]

Analogous to the uncertainty intervals that can clearly be seen around the centiles in the bottom right plot of figure 1 here.

For HBR, where the centile itself is taken to be the mean of the MCMC samples, the width of the uncertainty interval can be taken to be the variance. I.e., changing 'mean' here to 'variance'