I guess we all agree that an observation tries to capture the presence of a species in space and time, thus the taxonomic focus of an observation is a single species/taxon (rather than multiple species shown in some photos) as long as iNat’s data model isn’t fundamentally changed.
What we’re discussing here are biological and other attributes associated with the taxon-location-date core, and this in my view should allow annotation with all attributes that can be gleaned from the presented evidence. I suggest to open Annotations (and Observation Fields where applicable) for mixed categories or multiple states, and an updated definition of what an observation is.