The balloon trick is clever.

All that is SO much more work than rigging a proper set of cables, and with the servo in the float you'll have the servo wires looking like a scale set of heavy-duty jumper cables hanging out of the plane.

I have used Sullivan GoldNRods, but have never found that they are precise when bending much. They are great when run nearly straight, and the thing about temperature instability is only a ssymptom of incorrect installation. The 180-degree bend has lost motion and a lot of friction which causes poor centering, so the water rudders only point in the same direction as the air rudders by coincidence. ANd a big loop of goldnrod hanging out of the side of the plane is even less aesthetically appealing than the servo leads. If you find it necessary to use goldnrod, it works much better to run a nearly straight assembly from the nose-wheel tiller bar back to the water rudder arm.