Once you notice where a dead subpixel is though, your eye goes straight to it every time it's visible.
I wonder if this is actually true, the whole "more pixels than photosensitive cells" thing sounds a bit shaky.
Second, the number of rods and cones has nothing to do with it.
The "retina" argument was about having smaller than discernible angular pixel sizes, which is true for the majority of people and normal viewing distances.
Your conclusion does not follow from your premise.
Just because we can't identify individual pixels does not mean a single pixel has no influence. It shapes the overall picture.
If a single dead pixel is not visible at all, why have that pixel there in the first place -- working or not?
I just tested with my iPhone 6 (a 1334x750 px white picture with a single black pixel in the middle of the picture), and the single black pixel is definitely visible. A screenshot of the iPhone showing the picture in question confirms it's a single pixel on the screen.