The distance from the wing to tail is very short and the tail area is not that big a percentage of the wing. Both of these factors will gang up to produce a very small tail volume coefficient. And that will produce the need for the CG to be located much further ahead than most models.

It's not that your tail is stalling, unless you've got the CG lcoated right at the leading edge, but that you're balancing the model at or behind the neutral point so it's incredably touchy. I suspect that the thrust of the engine is producing a little nose down pitch that you're compensating for with some up trim so it fakes a condition where under power the CG FEELS like it's further ahead and so the model is stable. But once you power back that's gone and it's just a glider.

There's links in the sticky thread at the top of the thread titles in the Aerodynamics forum to some online CG calculators. You input all the factors I mentioned about the TVC along with span and chord and it crunches a bunch of numbers and tells you were the wing's MAC is located and where the CG should be located for the percentage stability margin you request in one of the boxes. Take a few minutes to run the numbers and see what comes out. Run it twice with stability values of 3 and 10 to get a back and forth range of acceptable CG locations. A stability factor of 10(%) is darn close to what a trainer flies like and 3 would be a fairly neutral handling sporty model

EDIT- Here's a direct link to the calculator
http://www.geistware.com/rcmodeling/cg_super_calc.htm

Also I misremembered what stability values you should input. Seems like 5 is about as low as you want to go.

There's also another issue with the U2. The fuselage is very long for the aerodynamic length of the tail. This fact along with the fact that it's a scale model with a fairly heavy engine in the tail is not going to help with the pitch damping in flight. So you're going to want to use an even more forward CG to aid this factor. And on top of all this you're using flaps on a fairly wide wing with a model that has a very low TVC. I strongly suspect you're getting some pitch down from the flaps at the same time that they are lifting. Again this calls for a little more forward a CG location to aid in dealing with that along with setting up some mix into the elevator so dropping the flaps raises or lowers the elevator as required to compensate for any nose up or down pitching. It should also go without saying that if your fuel tank is ahead of the wing that the model must be balanced with an empty tank. Otherwise the fuel load burning away will again shift the CG back too far.

I just wanted to say that I've seen a few models fly that had too rearward a CG and I've even flown one such model. Your videos are very consistent with how such models fly. So use that online calculator to figure out where your CG needs to be and re-trim the model to suit. You can ignore the wheel weight distribution bits. That's for folks using the calculator on full size aircraft.