In the PTC Help Center there is written that if you have redundancies, the total load doesn't vary but the particular loads between the constraints that have the overlapping, don't have the correct repartition of the load.
The example proposed was a door with two pins instead of only one: there are two bodies (ground and door) and the loads between them is correct; what is incorrect is how much load is taken by the first and the second one.
Since a slot has 2 doc (degrees of constraint), one along the tangent and one along the normal of the curve, this means that along this directions the slot transmits the forces and we could have a motion or a reaction depending if the successive body can moves or not into those directions.
Since the curve is a 3D curve, the slot transmits to the disc a force that, speaking in the coordinate system of the disc, has components in X,Y (and we have the torque transmitted for the rotation) and Z, along the axis of the disc. This is proved by the fact that if you replace the disc's pin with a cylinder, the disc starts move up and down.
So the pins on cam and disc are necessary but, with the slot, they cause a redundancy of 1. I don't understand this one which overlapped constrain is due and, consequently, which reactions are divided wrong.
My idea is that the redundancy is "inside the slot" between the load repartition between tangent and radial component. On the disc you don't know if the motion is greater caused by the X;Y o Z components of the slot but, the combination of those three, projected on the disc plane, gives the right force and the right torque.
If this is true, I can easily accept to have a redundancy because it doesn't affect my motion analysis.