In my idea, there would be a bearing, perhaps a plain bearing, between the rod/tube + … + (vane) and the rotor.
I think the rod is mandatory. You could try it, what happens when you put a rotor on an inclined rope? I think it would just slide up and down the rope with no way to transfer the pull. You could put stoppers on the rope, but then you’ve already introduced the rod, really. The rotation would also wear down the rope and be a source of friction.
You’ve got the rope and the rotor, you could try this out now.
To make the top rotors pull more upwards.
… so above I’ve put a bend in the rod. You could also make that a hinge, which would give you some control in the direction the rotor would be pulling. You’d need that hinge to always be oriented correctly, otherwise you have a chance of the rotor pulling more or less left or right or down instead of more or less up. What should that hinge do and what are possible inputs to it? Other than actively controlling it from the ground [1], there’s perhaps relative wind speed, tether tension, tether angle, rotor pull, gravity. You could try to use some of those inputs to control the hinge. You’d want to control tether angle I think.
Or instead of or in addition to adding a hinge, you could make some bends in the rod that control in which direction the rotor is pulling based on how strong it is pulling, the rotor would be sliding up and down the bendy rod.
[1] If you’re going to be actively controlling it from the ground, you could make a program that keeps the (graph of the) tether curve within what you want. Every hinge could potentially receive different inputs.