Heh, now you open this Wardley Maps Pandora Box :-). If you say "landscape" (I agree it's a better term than "reality"), which reflects structure, you will probably also need to say "climate", aka dynamics (btw not addressed sufficient in the SDDD lingo). While the both factors are independent from the "observer" (us) they are largely interacted with through subjective perceptions and while might be accurate enough do not have to be precise (more often they are not). Maps (models, Bounded Contexts), on the other hand, reflect the observer's attempt to formalize these perceptions with agreeable precision for the sake of efficiency, but are at the permanent risk of losing accuracy with regard to the original landscape/climate.
That effectively means that Subdomain analysis and Bounded Context modeling (could we say it this way) are in a constant dynamic shift flux.
I could virtually imagine how you pull out the Liminal Cynefin diagram to illustrate this dynamics. I always liked your approach of pulling together multiple complimentary disciplines to get some useful stuff done. Keep on going.