Feature: exponential delays

[seabbs]

This PR closes #83 by adding support for exponential delays. It has no impact on default functionality. Checking the exponential delays posteriors they look reasonable. Checking posterior predictions indicates that performance is slightly worse than with the log-normal (at least on this very limited test set). The major issue seems to be capturing the first day of reporting which I find a little surprising.

[sbfnk]

Looking good to me. Would it be worth reserving dist == 0 for a model option without parametric delay distribution at all, i.e. the nonparametric components only?

Minor point: should update the comment in l.33 of inst/stan/epinowcast.stan.

[seabbs]

Thanks for the review. Yeah, I considered this and decided against it initially. Reconsidering I think you are right, retooling this now and adding the ability to turn the parametric model on/off at the same time (though without putting anything else in so this will just break if accessed).

Note bumped from ctb -> aut based on contributions: 597e5cb

Comment fixed in: 2f254b3

[seabbs]

I've updated this so that dist = 0 corresponds with no parametric hazard model. At the moment that gives something fairly silly as the non-parametric reference day model is not in place but will be useful for the future.

As part of these changes I've done a little bit of internal refactoring to make things easier (@adrian-lison
I will add this in to #64 before we catch up on it).

[codecov]

Codecov Report

Merging #84 (2ae1a32) into develop (6390031) will decrease coverage by 0.50%.
The diff coverage is 27.77%.

❗ Current head 2ae1a32 differs from pull request most recent head 10f49c6. Consider uploading reports for the commit 10f49c6 to get more accurate results

@@             Coverage Diff             @@
##           develop      #84      +/-   ##
===========================================
- Coverage    55.65%   55.14%   -0.51%     
===========================================
  Files           12       12              
  Lines          911      923      +12     
===========================================
+ Hits           507      509       +2     
- Misses         404      414      +10     

Impacted Files	Coverage Δ
R/model-tools.R	0.00% <0.00%> (ø)
R/model.R	22.54% <62.50%> (+0.51%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 6390031...10f49c6. Read the comment docs.

[seabbs]

Merging but we should circle back with a check beforedevelop -> main