first obvious issue is bearings and space between them. I do not know the force you need to bend a rod or tube compared to the ball bearing options. Journal bearing can carry higher loads and a good lower cost and size bearing. Speeds are low so they will have a long life. Next issue is dealing with the cantilever of the rollers. Do a force balance. Image the shaft is a beam with the back bearing a pivot point. Apply the force for bending at the point on the beam where the roller is. Then calculate the balance force for the bearing between the pivot and force. Force middle bearing = force roller x distance pivot to force / distance pivot to middle bearing. Then balance forces pivot force = roller force + middle bearing force.
As you see the middle bearing is the biggest force. My first change would be to increase the distance between the middle and pivot bearing. and keep roller close to the middle bearing. A better solution is to put the roller between the bearing which may not make sense for usage.
The pivot load if the spacing is large is small. Thus the back plate may not need to be bigger and you can mount driving chain sprockets on the shafts of the to none moving shafts. The moving roller is a little harder to space out the bearing but I suggest doing it . The location of the screw to move the shaft would change to the point where if this was the pivot point in the above calculation would be balance. Dist from mid bearing (above) to screw = force pivot bearing(above) x space between bearings / (force pivot bearing + force mid bearing) The screw will help keep the shaft level and the torque to rotate it will be less in the force direction. Not considering forces from driving the rollers in this analysis.
