Update discrete_reporting_prob

[seabbs]

After discussion (between @adrian-lison and myself) we concluded that the discrete reporting probability should be normalised to 1 by putting 1 minus the cumulative probability density on the last permissible max delay rather than the current default of normalising. This should be added as an option in the stan code but hard coded in the actual model as no other use case makes sense currently.

Before merging we need to fit against the example and README to check performance.

https://github.com/epiforecasts/epinowcast/blob/a5616b15857d790eced0f1289b498d42fa0ca6d7/inst/stan/functions/discretised_reporting_prob.stan#L3

[seabbs]

Done in #113