Here's a glimpse on what's happening in the background, maybe it can shed a little light on these answers in regard to the resolution options. The survey asks the user to fill out their monitor's native resolution then asks if the game runs full screen without stretching. Enhanced Steam compiles these answer pairs into a dataset and then performs a Bayesian analysis on that dataset to come up with a "confidence score" for each answer pair. The answer pair with the highest confidence score is the one that gets displayed here.
So in the case of GTA V and by looking at the back-end data, I see that a few people with 4k reported it didn't play full screen (this is probably incorrect, or maybe they just didn't read the instructions, it's hard to say with things like this that are crowd sourced) but everyone that reported a 1080p monitor also reported that it played full screen without stretching. There were also a lot more people that reported 1080p vs 4k, so the system's confidence score in 1080p being the right answer was chosen.
I think the confusion may be due to the wording here, which I may update. I think it might be more accurate if it said it appears to support at least 1920x1080 resolution.