i know this wasn't the intention, but i think this is where things got confusing:
the moment you say "amp in the room" it sort of does become cab vs FRFR. most people will define "amp in the room" specifically as a real guitar amp/cab bouncing around the room. there really isn't a way to do that with an IR or FRFR speakers. so you could say that, by definition, the title of the thread was somewhat "real cab vs FRFR" at least in theory.
maybe that was already cleared up and understood, but that comment just stuck out to me.
all that said, personally, i experience "amp in the room" with my FRFR speaker when it's loud enough to bounce around the room and it isn't facing my ears. in other words, use full modeling, use an FRFR, but just put that FRFR speaker in the same physical placement and volume as a real guitar amp - facing straight forward to your ankles and loud enough to bounce around the room.
it may not be exactly the same thing, but i personally don't feel much different about that.
the thing about direct/modeling is that you can use headphones, or speakers that angle toward your ears, or many other ways other than how a real amp/cab is used. guitar amps are notorious for being "too loud" because they are directional and you need certain volumes to create certain tones. but with modeling, we don't do that (because we don't need to), but some of the qualities are strict physics and nothing to do with modeling vs non-modeling.
try this - spin your FRFR speaker away from you, turn it up - are you getting the room sound now? and no it can't be in a sound-treated room that doesn't bounce sound