@User: You'd have to read a lot of background material about magic to understand why it does some things the way it does. But yes, that's right; magic wants to divide up "diff" between N and P and show them as separate layers. This makes it much easier to tell what's going on in a circuit because the N and P devices are immediately recognizable from any distance. Magic also divides "diff" and "tap" into separate types, which a lot of processes don't do, but SkyWater already does that in its GDS layer assignments. There are a lot of "ID marker layers" in the SkyWater process, mostly for LVS and DRC purposes. "areaid.sc" is an ID marker layer for standard cells. A few of the DRC rules are less strict inside standard cells than elsewhere (the main one being the source/drain contact to gate distance), and so the ID layer helps the DRC engine determine what rules to use.

Thanks @Tim Edwards for detailed explanation. I guess the "diff" and "tap" are represented by "ndiffusion" and "nsubstratendiff" respectively in magic.

@shbo : Yes, that's right, although there are various aliases for each layer, so "ptap" and "ntap" also work for the tap layers.