Create forecasting functions

[apulsipher]

We will need forecasting functions that use epiworldR to generate the forecast. These might be embedded first in the quarto document, but ideally should be separate and generally useable (possibly even part of the epiworldR package or a separate R package).

The forecasting functions should sample from the posterior distribution of the LFMCMC simulation and run the model for each sample parameter set (rather than on the second quantile like we do currently). This will make our forecast a distribution rather than a single point prediction.