implement repeat random sampling of GI from family when daily reports from aggregates

[papsti]

this may be slow as it involves re-running the JAGS model several times. the current approach is just to do one run with the mean GI. a middle-ground between these two approaches would be to sample the mean GI, some lower/upper bounds (using parameters taken from some quantiles of the parameter distributions). this would only involve three runs of the JAGS model.

[papsti]

implement this as an option---if one wants to resample they can, but by default just use mean