I am moving countries again. This is a final test assembly before I toss the bulky stuff and pack everything I own into a single suitcase. The test did not go as I was hoping, and there are still a lot of things to refine. 

The chinese splitters I was using I believe had a short, which causes some magic smoke with the controllers. I did get some useful results though, I'll do a quick summary below.

• While I thought at 50% power draw the TEC 12715's would be sufficient for cooling, I feel one side getting warm to the touch and the other is not noticeably chilling the water at all. I will have to do another test one at a time to make sure it was not the short causing this. Otherwise, I will switch back to the 12706's if the 12715's need considerably more power to effectively cool than I thought. The capacitors of the MD13s cytron boards was a bit hotter than I'd like. Increasing the power draw caused the splitter to melt and short I think.
• The nice pump I ordered has to be in a specific sideways orientation or gravity fed by the reservoir to work effectively. The 3D printed reservoir had a leak with the front plate. It needs to be a more solid material or printed as one solid piece to avoid leaks. I made another reservoir with a screw top printed as one piece, and it did not leak badly. Horizontal placement of components makes this complicated.
• I think it is not worth it to try and continue to iterate on the flat design. The tower is better physics-wise and will let me gravity feed the pump, which is far simpler user experience.
• It sucks to route the wires
• PSU above the water was important (I got shocked anyways)
• Screw top does leak a bit, the 3D print design needs to be better optimized so you don't have to screw it so tight to get a pressure seal. 
• The 3d printed connector on the back leaked way too much, it needs to be coated in epoxy. 
• While the black hose looks cool, you aren't able to see the flowing water for testing. This messed me up a bit. 
• I think the mounting bracket for the heatsinks will need to be redone to better isolate the copper heat block. Even though it looks so sexy, I think I will have to kill it.

• Switched from aluminum to copper because you shouldn't mix those I guess. The fittings are coated copper or brass.
• Some issues with the pump speed not changing as expected. Runs at 100% which is too loud. I need to figure out how to run it lower. Circulation speed was pretty good. 

So I think in conclusion, 8Sleep pod is pretty close to the optimal layout already. There's a reason they aren't laid horizontally. I think the V2 will be a cuboid. See you in a few months when I have (hopefully) a new workshop. I do not know where I am going yet, just that I am leaving in three weeks. Might be longer.