investigate priors in smoothModel by looking at posterior distributions of non-existent sites

[famulare]

I've tuned the prior so far based on median smoothing, which isn't ideal. We can also get a good sense of priors by looking at the posteriors in regions with no data. I saw an example of this with self-test prevalence extrapolation that should be thought about more.