We had a discussion about this, maybe even on this forum. The guy I was arguing with sent me a video of the exact setup we are talking about, and I superimposed predicted motion over that video.
Here is the resulting video.
The red dot on the arm is the center of mass between the slide and the rotating mass. The green line is predicted motion of center of mass based on assumption of no friction. As you can see, red dot deviates a bit in the end, most likely due to friction in the slide.
Here is what's going on. While the rotating mass is on the left, the tension pulls the slide to the left, and it's prevented from moving that way. So it works as a stationary pivot. As soon as the arm passes the low point, the tension starts to pull to the right. From that moment on, the slide is free to move.
Now, the slide + rotating mass can be viewed as a closed system. There are no sources of horizontal force, so horizontal velocity of center of mass must remain constant, while vertical velocity is just given by rotation of the arm around the pivot. That's the green line, and that's the motion of the red dot.
The angular momentum of the arm around the pivot point is conserved, since there is no torque about the pivot. That means angular velocity of the arm is constant. Knowing these facts, you should be able to project the motion of both the slide and the rotating mass. Knowing velocities at each instant of time, you can derive accelerations, and therefore, forces acting on the slide and the rotating mass. Naturally, you'll find these to be equal and opposite in the horizontal direction. In vertical direction, you also have to account for the normal force provided by the rail.