Improving prior specification

[seabbs]

At the moment it is hard to set most priors though it can be done via the data frames output by each module and the priors argument to epinowcast for some parameters

There are two options open to improving this. One is to adopt a prior system from somewhere else, for example, brms or rstanarm and integrate it into our code (as done in epidemia). The other option is to expand the current system so as to also include model effects and random effects. The latter could also be a partial/temporary step along the roadmap of the former.

When doing this work its important to keep in mind that a key use case is setting constant priors (as discussed in #33).