Just thinking out loud… In order to create a solid genus atlas, one would already have to “know” what all of the species/subspecies atlases look like for that genus. So it seems like that is something that can/should be aggregated from species atlases by the system at the time of a split, without having to store a separate genus-level atlas (and having to keep it coordinated with its component species, etc.). And that could happen just as well for any other rank above species.
I’m not sure what the best option is either, though I’ve definitely used that workaround to create taxon splits for genera. Doing the mapping from species up would be very tedious or impossible in some cases. (It’d be interesting to see if many people are even creating atlases anymore? I mostly stopped, aside from when it relates to taxon splits, since it was too frustrating to maintain them.)
I’ve only created them in anticipation of a split, except for maybe a handful of taxa whose distributions are of special interest to me.
I atlas any new mammal I add, even if it’s de novo and not a split.