Issue 4: Non-parametric baseline hazard

[seabbs]

Description

This PR closes #4.

It adds a non-parametric baseline hazard model to allow the specification of @FelixGuenther original model or to get close to specifying @adrian-lison model. The current implementation does some non-optimal index hacks to get this working that I think should be resolved in a future PR (as indexing could in general be simplified). @kgostic this is likely useful for some of our applications but is obviously still a work in progress.

Work done:

• Implement an example script with desired functionality
• Add support to the stan code
• Add support to enw_reference()
• Remove all built-in warnings and messages about the non-parametric model
• Get tests updated and fixed
• Add a test for the new functionality in enw_reference()
• Add tests for new functionality in epinowcast().
• Update snapshot tests
• Update cached example data
• Open an issue for improving indexing. See Improve stan code indexing #320
• Open an issue for discussion about updating the default model. See Update the default model #319
• Open an issue for handling hazards more efficiently for the max delay (as not set 1 in the non-parametric model and still allowed to be parameterised). See Improve handling of the maximum delay when only a non-parametric hazard reference date hazard is used #321.
• Open an issue to check synthetic recovery of non-parametric reporting delays. See Add synthetic validation tests for the non-parametric and parametric reference models #322

Checklist

• My PR is based on a package issue and I have explicitly linked it.
• I have included the target issue or issues in the PR title in the for Issue(s) issue-numbers: PR title
• I have read the contribution guidelines.
• I have tested my changes locally.
• I have added or updated unit tests where necessary.
• I have updated the documentation if required.
• My code follows the established coding standards.
• I have added a news item linked to this PR.
• I have updated the package development version by one increment in both NEWS.md and the DESCRIPTION.
• I have reviewed CI checks for this PR and addressed them as far as I am able.

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 3881763 is merged into main:
These benchmarks are based on package examples which are available here. Further explanation regarding interpretation and methodology can be found in the documentation of touchstone.

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 2734ea8 is merged into main:
These benchmarks are based on package examples which are available here. Further explanation regarding interpretation and methodology can be found in the documentation of touchstone.

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 2734ea8 is merged into main:
These benchmarks are based on package examples which are available here. Further explanation regarding interpretation and methodology can be found in the documentation of touchstone.

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 2734ea8 is merged into main:
These benchmarks are based on package examples which are available here. Further explanation regarding interpretation and methodology can be found in the documentation of touchstone.

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 2734ea8 is merged into main:
These benchmarks are based on package examples which are available here. Further explanation regarding interpretation and methodology can be found in the documentation of touchstone.

[codecov]

Codecov Report

Merging #313 (ee533c3) into main (2734ea8) will increase coverage by 0.17%.
The diff coverage is 100.00%.

❗ Current head ee533c3 differs from pull request most recent head 8307946. Consider uploading reports for the commit 8307946 to get more accurate results

@@            Coverage Diff             @@
##             main     #313      +/-   ##
==========================================
+ Coverage   96.61%   96.78%   +0.17%     
==========================================
  Files          15       15              
  Lines        1771     1804      +33     
==========================================
+ Hits         1711     1746      +35     
+ Misses         60       58       -2     

Files Changed	Coverage Δ
R/model-modules.R	99.50% <100.00%> (+0.58%)	⬆️

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 2734ea8 is merged into main:
These benchmarks are based on package examples which are available here. Further explanation regarding interpretation and methodology can be found in the documentation of touchstone.

[seabbs]

LGTM (self-review)

[seabbs]

The lintr CI check is failing when trying to install dependencies (specifically vdiffr) for unclear reasons that seem unrelated to this PR.

I've run lintr::lint_package() locally and fixed any issues that were flagged.

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 2734ea8 is merged into main:
These benchmarks are based on package examples which are available here. Further explanation regarding interpretation and methodology can be found in the documentation of touchstone.

[seabbs]

LGTM (self-review 2)

[seabbs]

I've done two rounds of self-review and am happy this is implemented approximately correctly (enough to be a rolling release candidate). Will work on #322 as a priority to help give extra confidence.

Given lack of review activity (which is fair enough) going to merge this but always happy to have a post merge review.