I really like the screw approach to stub turnout control. One of those ideas that makes you say “why didn’t I think of that.” It’s a derivation of the “choke cable” approach, but sounds like it would hold the rails in place much more reliably, and likely would make it easier to perfectly align the rails in the first place. Of course it means you’d have to eyeball the turnout every time for alignment, but that could be an advantage as I suspect the nails (or whatever) one might use to stop the rails would be miserable to get perfectly located, and might just be subject to moving around over time.
The moveable rails must, of course, not be attached to the ties for some distance. On the prototype the typical approach was to install a number of rods between the rails to keep everything in gauge, and there apparently were usually metal plates on the tie tops for the rails to slide on. I don’t think there was anything to keep the rails down against the ties except for gravity and the weight of stuff rolling over them. I cannot think of any practical model version of these rods. Plastic (anywhere near scale) wouldn’t be stiff enough and metal wire would work only if we didn’t have to keep the rails electrically isolated. I am concerned with keeping the movable (model) rails aligned vertically as well as horizontally, and see that as a potential problem. Model point turnouts have the advantage over stub turnouts, of the tie rod sliding between the rails and the roadbed which keeps the points vertically positioned and level with each other.
I remember one approach in a model magazine many years ago, where flex track was used and the ties were left attached to the rails --- allowed to move with the rails. At first I thought that a great solution, and was what I used in my experiment with modifying Atlas turnouts. Worked great, and although it of course wasn’t prototypical, looked pretty good too. That is until I began to wonder how I would ballast the roadbed in that area. Not only would the ties have to stay loose, the ballast could not get in the way of the ties moving. That would be such a hassle, it caused me to discard the whole idea.
Now I’m thinking that the solution might be to use a wide tie rod (or whatever you want to call the thing that holds the end of the moveable rails in gauge AND moves them.) Wide enough that it slides under the ends of the fixed rails as it moves the moveable rails. It would take some thinking to keep it from looking clunky. Perhaps it could be made to look like two ties with some ballast in between, but in fact the whole thing would move. I know I’m not explaining this well, but my scanner is down so I cannot show you a sketch of what I mean.
BillS