Fix broken quantiles (#88)

[jstockwin]

Closes #88

[jstockwin]

@zkamvar Would you mind taking a look at this when you get the chance?

[zkamvar]

This looks good to me, except now the tests need to be adjusted to reflect the true quantiles. Would you mind fixing the binary files in test/expected_output?

What you cannot fix: the plots are failing, but that's due to a bug in the new version of ggplot2 (tidyverse/ggplot2#3873) that caused incidence plots to fail (reconhub/incidence#119).

The current version of incidence on GitHub gets around this by centering the dates on the bars, but has a shifting effect.

[zkamvar]

This looks good to me. There really is nothing we can do about the incidence problem until ggplot2 gets fixed.

[jstockwin]

Thanks. @zkamvar. Do you think we should wait for the issue to be resolved, or just merge with failing tests?

Currently I think I can't merge due to requested changes ("Merging can be performed automatically once the requested changes are addressed."), and probably it will say the same for failing tests... Edit: Github was being weird, it now says approved, but I can't merge due to failing tests.

[zkamvar]

Thanks. @zkamvar. Do you think we should wait for the issue to be resolved, or just merge with failing tests?

Currently I think I can't merge due to requested changes ("Merging can be performed automatically once the requested changes are addressed."), and probably it will say the same for failing tests... Edit: Github was being weird, it now says approved, but I can't merge due to failing tests.

I think now that there is a new version of incidence on CRAN, the issue should be resolved. I believe the solution is to open R in the project and use vdiffr::manage_cases() and just make sure that they look generally okay (again, the bars will be shifted to the left a bit because of the nasty hack we had to do to get things to work properly [although I supposed the original solution itself was a nasty hack, so we are negating the nasty]).

[jstockwin]

@zkamvar I've updated the figures using vdiffr::manage_cases(). The bars appear in the same place, but the x-axis markers have moved to the right (which I guess is the same as the bars moving to the left and the whole thing then being moved right).

Hopefully the tests pass now...

[jstockwin]

@zkamvar Tests passed (here), but on my screen at least github is still saying "Waiting for status to be reported". Not sure if it's a Travis problem or a GitHub problem... hopefully it sorts itself out :/

[jstockwin]

@zkamvar I have restarted the travis build, since something clearly went wrong. I can now see that GitHub knows it's running, so this time everything should actually go green. Not sure what happened there...

Are you happy to merge once green? Do you know how to get an updated version on CRAN?

[jstockwin]

Also I've just pushed a new commit because I forgot to update NEWS.md...

[zkamvar]

LGTM! Thank you ^_^