On one of my many long-term tests, I rode the 'Q' rings for a couple of seasons. I found them quite good. I logged quite a good set of data and could compare vs round. The manufacturers claims are a bit steep, but I think I can say that there is an advantage for certain applications (short, high-powered efforts).
With respect to the negatives, I think they don't feel quite as good as round rings (which is one of the reasons I am back on round).
Don't listen to people who say that you will have shifting problems or that the set-up is difficult. That is baloney. They are just as easy to set-up as round rings and they shift just fine. They are not as good as the 'factory' rings with their marvelous tooth shapes, dimensions and pins, but they work just fine. If you run Di2 or EPS, they are slightly flexy, but not bad.
Because of the way that they feel (different than round), I would not recommend mixing rings on one set. Even if you are changing between different bikes, you notice the rings.
I think I would agree with you. They felt really good on the trainer/rollers or in a fast paceline. I can't say they made a big difference in terms of power production or anything and the data showed no real differences in my abilities on either. I ended up with a 4 bolt PM and had to switch off of them abruptly during race season and didn't really notice.
I had Q-rings with Ultegra di2, Campy Record/SR11, and DA9000 and they shifted really well actually. A hair less crisp than my perfectly adjusted, newly-cabled 9000 rings but not enough to really care. I don't know about QXL. I used OSys for a month at one point and shifting was fine with 7900. Took a bit to figure out the cage angle and spacing, but it wasn't as bad as some people have claimed.
I would never mix rings. The benefit seems to come from a specific effort/cadence, which you will experience on all types of challenging terrain.
Don't take me too seriously. GramzStrava