z22 1.2.0

Major changes:

Perform full joins rather than left joins when multiple categories are provided.
- Earlier versions can severely underreport cell occurrences when downloading multiple categories so it is recommended to depend on version 1.2.0.
normalize = TRUE now computes shares based on the sum of subtotals instead of population totals as suggested by the census methods paper.
Several corrupt category codes that were not used by the census data were removed. This includes:
- floor_space: code 99
- dwelling_rooms: code 99
- dwelling_occupancy: codes 99, 1, 2, and 5
Added a category specification for dwelling_building_dwellings.

Minor changes:

Cache files are cleaned up when a download fails.
z22_decode() now supports a year argument to account for category changes between censuses.
NAs are replaced with zeroes for shares when all_cells = TRUE.

z22 1.1.4

Replace rasterize and as_sf arguments with a single as argument that takes "df", "sf", or "raster"
Fix z22_inspire_generate() for the long format by not truncating trailing zeroes
Rename legacy to short for INSPIRE identifiers. Short and long are more descriptive than the previous naming scheme
Add an argument llc to z22_inspire_generate() that specifies whether the coordinate pairs are already lower-left corners or must be adjusted accordingly
Add an argument meta to z22_inspire_extract() that controls whether to return metadata like CRS and resolution.
Add additional checks to z22_inspire_generate()

change: z22_data() and z22_grid() now return a SpatRaster instead of a SpatRasterDataset
fix: a bug where z22_grid() could not rasterize grids
fix: shift grids produced by z22_grid() by half a cell to keep grid alignments consistent
fix: allow cache updates for auxiliary population counts when normalize = TRUE