VR hardware is great step-up from the common gaming setups, where you only have to render a single scene @ 30-60 FPS and display it on a monitor @ >1080p. VR is much more involved than that. Not only do you need to render the scene twice, but you have to do it at a resolution >4k per eye (which probably isn't even enough to satisfy all users), and on top of you should do it in <20ms or something in that order, to be sure nobody gets sick. _including_ position tracking.
The rest of your points can be solved, once we can mass-produce hardware that meets the specs.
I've heard perfection needs less than 7ms latency. That's over 140 FPS, possibly higher if you have any kind of double buffering.
I'm sure there's someone working on a headset that tracks eye movements and adjusts the rendered resolution of the image on the fly. The human eye has something like 4-5 megapixels worth of photo receptors, but the absolute majority of them are located in a relatively tiny area of your FOV where your eyes focus on.