fix: manually do smoothing in jhu_csse_daily_subset, regenerate

[dshemetov]

closes #251

changes

• instead of using API smoothed data (which is out of date and has discrepancies), manually smooth using epi_slide
• completes this column product (cases, deaths, case_rate, death_rate) x (smoothed, not smoothed) instead of doing just a subset

note that archive_cases_dv_subset.R also uses the unreliable confirmed_7dav_incidence_prop data, but it gets the full archive of data and fixing that is much harder.

[brookslogan]

The rda looks like it's probably good, but it doesn't match what's generated by the current R script on this branch. I think you can probably just back out the changes to the R script and we're good because I think the API's remaining errors are only in older as_of data, not latest data.

but also

• todo: "fix doc bug to note that Delphi did 7dav-ing" [this has been long-resolved?]

• If you think the backing out changes thing works, then just addressing that last TODO is probably all that's left. If you think the epi_slide approach is needed, then I think re-review would be good to get another sanity check.

[brookslogan]

Also

• todo: add NEWS.md entry

[dshemetov]

not sure what "fix doc bug to note that Delphi did 7dav-ing" means. where in the docs?

sounds good, i backed out my changes, moved lines around in generation file, and regenerated.

[brookslogan]

Data set docs are in R/data.R; this should be the first one (attaching roxygen docs to "jhu_csse_daily_subset" (rather than e.g. a function declaration) documents the topic jhu_csse_daily_subset).