I've never run into one acting "just like" this one.
I do have some training materials here and if I get time I can study them a bit.
The top operates in a series of steps. It can't go on to the next step without verification by various micro-switches and sensors that the previous operation has been completed. Simple enough so far, right?
So for example we know that the windows need to go down, including the 1/4 windows, and it is doing that. We know (I believe) that the top unlatches above the windshield and begins to come back, and that the trunk lid is opening (pivoting rearwards). These operations all have switches and sensors to indicate to the control module that "Yes, this is done, proceed with next operation, we're good to go". It seems like what you are saying is it gets about as far as I've described and then just stops. There is this computer the tech uses, what they refer to as "sds" (star diagnosis system) that can monitor all these switches and tell if the control module is getting "the green light" to proceed from all the sensors and switches.
The one I personally would scrutinize the most is the one that tells the control module that the trunk is all the way back. No way in hell the control module will try setting the roof into the trunk if it isn't sure that the trunk lid is out of the way. CRUNCH!!!
What may be happening is the pump may continue to run (in case the trunk lid is just moving slower than it should) while in fact it IS all the way back (poor switch, not sending a signal) and is overheating the hydraulic fluid and possibly even causing the rams to seep. This whole thing with topping up the fluid, boy I sure hope the rams aren't junk now. My first recommendation for the time being is to NOT hold the switch and run the pump, this isn't proving anything. Just making things worse.
They've gotta go back to square one on this thing I guess. Get SDS hooked up, look at the signals they are getting and find out IF they are getting the proper signals and if it's not working right, sure, then maybe suspect the hydraulic unit or control module. I think the whole reason that the control module thinks it has a jammed hydraulic ram is that the switch telling the control unit the decklid is back all the way is bad or misadjusted. The only way the control module knows that a part of the top is where it should be is by the switches, it's not THAT SMART of a unit that it just "knows". They need to clear codes, then watch the switches and look at where everything is when the top stops working, they should be able to figure out RIGHT THERE where the problem is. Unless at that point there is fluid gushing out of a ram, but I bet it's not.
Gonna try another dealer, or print this out and show it to them, or what?