Currently, r.resamp.stats operates at two different resolutions
-- the one obtained from the current region (the resolution of
the destination raster) and the one obtained from the raster the
module operates on (the source resolution.)
Does it sound sensible to generalize the concept and allow for
arbitrary region to be specified for the source resolution (so
that the source raster is effectively downsampled prior to
applying the aggregation function)? May it make sense to add
explicit support for the secondary region at libgis level?