Fix prob_arx misalignment in this repo as well, or remove buggy vignette( section)s

[brookslogan]

Fix cmu-delphi/epipredict#50 in this repo, or remove the buggy vignette( sections)s so they aren't copied.

[dshemetov]

Closing as outdated.

[brookslogan]

I still see prob_arx in archive.Rmd and it looks like it has the same issues; perhaps the fix is in some open PR that would move us to using arx_forecaster that attaches an appropriate target_end_date? "Appropriate" here meaning actually representing what the model is targeting.

  dat$y <- lead(y, n = ahead) # no reporting latency adjustment to make .data$version + ahead below make sense
...
      mutate(target_date = .data$version + ahead, as_of = TRUE) %>%
...
      mutate(target_date = .data$time_value + ahead, as_of = FALSE) %>%

(Also, I thought we were planning to replace these epi_slide things with epix_slide on an archive prepared with snapshot %>% mutate(version = time_value + fixed_latency) %>% as_epi_archive(), though unless fixed_latency holds true for all initial reports or we do manual latency adjustment, this means that the "appropriate" target end date labeling will show that the naive & version-aware methods are targeting different dates.)

[brookslogan]

This was already resolved here via removing this section and pointing to an epipredict vignette.