so, it's only about the window regulator on the driver's door, the others are fine and the central locking works too.
Generally speaking: the Cupra probably has the same tilt/pull switches (four) in the driver's door. If you press a switch briefly, the corresponding window goes down a bit, just as you describe. If you press the button a little longer (the window is not all the way down yet), the "after-run" kicks in
This would mean you haven't activated the "after-run." Maybe just a "mü" not right now. Perhaps the window regulator will catch up on the "after-run" the next time you press the button.
Specifically: you write
"When I want to raise it again, I press the switch to close." With the ROC, you have to pull the switch up to close the window, not press it.
Assumption/thought: if you press the switch again, the window logically goes down. But you want to close it and might therefore be holding the button down. Could it be that the window regulator consequently skips the lower stop point and goes back up? That shouldn't be possible, as that would be the "learning."
Or does the Cupra have different buttons? Then I'm unfortunately at a loss.
How does the window regulator behave during "comfort opening/closing" (with the key)? Does that work flawlessly? Then, in my opinion, at least the door control unit should be fine, so that "learning" the key and the central locking is unnecessary. Maybe only the switch is defective.
One more thought: does the Cupra possibly have a resistance detection when closing the windows? Could it be that at some point a resistance was detected and therefore the closing was interrupted? I could imagine something like that especially with the heavier acoustic glazing (the motor has to work harder).