find_hex_cell_size_for_target_cells() derives sensible cell_size_min/cell_size_max defaults from the study-area bounding box when those arguments are NULL.
New global option hexsmoothR.verbose (default TRUE) controls all informational console output via the internal hex_msg() helper.
Bug fixes
extract_raster_data() now reprojects the hex grid to each raster’s CRS before calling exactextractr::exact_extract(), fixing silent CRS mismatches with multi-raster inputs. The cropped study area is also rebuilt per raster.
extract_raster_data() returns the (possibly sampled) hex grid in the result list as hex_grid, and the “Sampled X from Y” log line is now correct.
get_utm_crs() correctly handles longitudes near 180 and auto-projects inputs to WGS84 before computing the centroid.
smooth_variables(): the raw element of the output now stores the original (unsmoothed) value of the centre cell, as documented.
C++ Rcpp::stop() messages no longer print literal %d placeholders.
compute_topology() order-N BFS rewritten using set operations; large grids and higher orders are noticeably faster.
Documentation
Vignette hexsmoothR-complete-guide has been rewritten end-to-end and is now driven by the bundled inst/extdata/default.tif sample raster.
Added NEWS.md, BugReports, and a second URL for the GitHub repo.
Internal
Test suite reorganised into focused files (test-create-grid.R, test-compute-topology.R, test-smooth-variables.R, test-utils.R, test-get-utm-crs.R, test-vignette-runs.R, test-vignette-workflow.R, test-n-order-smoothing.R).
useDynLib(hexsmoothR, .registration = TRUE) added so registered C++ symbols are correctly resolved.
