sle_adm3: regenerate the stored object so validTo is NA

[raubreywhite]

Fixes #48.

get_data(name = "sle_adm3") returned validTo = -1-11-30 in all 167 rows; the shapefile field is empty (NA). The stored .rda had been built with an older GDAL that wrote a sentinel for the empty date. This PR rebuilds data/sle_adm3.rda from inst/extdata/sle_adm3.zip with the existing data-raw/sle_adm3.R under sf 1.1.1 / GDAL 3.4.1, so the object equals sf::read_sf() on the zip in every column (names, 167 rows, classes, CRS, geometry). Old and new differ in that one column only.

A sweep of all 103 stored datasets (210 Date/POSIXct columns, plus the nested epinow, stars and phylo objects) found no other pre-1900 date; sle_hf, Blocksshape, Quartiersshape and gis_africa_countries carry no date columns.

Version 0.0.0.9013 and a NEWS entry. No code change.

Check after install:

x <- appliedepidata::get_data(name = "sle_adm3")
all(is.na(x$validTo))   # TRUE
nrow(x)                 # 167

Downstream: the Epidemiologist R Handbook's GIS chapter switch to get_data() (appliedepi/epirhandbook#446) waits on this, followed by a pin bump in aedockerpublic (appliedepi/aedockerpublic#2).

[Lfquezada]

@raubreywhite The fix worked, but only by accident: it was built once under an old GDAL version, and the data-raw/sle_adm3.R script itself didn't actually handle the sentinel (so re-running it on a newer GDAL (like on my computer) reproduced the original bug.

I patched data-raw/sle_adm3.R to explicitly convert any implausible pre-1900 date to NA, regardless of GDAL version, then regenerated data/sle_adm3.rda from that fixed script (the data itself didn't change, it's just now reproducible).

Bumped up the version of the data in the tableoftables.xlsx as well!

The ID now would be ebola_outbreak_sle_2014_shape_1_2

Tested on my computer and looks fine! @jarvisc1 Ill merge now and update the package to V0.0.0.9013