Alright, so... for drain-back, I'm thinking there has to be a way to admit air at the high point on the collector (after the circulator is shut down, of course) so that gravity can draw the water out and air into the collector. Later, when the air entry is closed and the circulator runs again, the air will be flushed out of the collector eventually into the water tank, and normally one would think that air would need to be vented out of the system -- either by "burping" a closed system, or by having the tank open to the atmosphere.
What if the tank were closed and only partially filled such that there's enough air held in the tank for filling the collector? A tube from this tank air space up to the top of the collector could allow air from the tank to rise into the collector while water from the collector falls back down into the tank. The system could remain sealed. This requires a third tube going to the collector, a valve, and a way to ensure that the air tube stays above water level inside the tank. Is a simple check valve good enough for controlling the drain-back? The pressure from the pump would have to close the valve so that water doesn't short-circuit around the collector instead of going through it.
What do you think? Go ahead and pick it apart for me.
Remember too that this only addresses the boiling problem; the collector still has to be made to withstand the high temperatures it'll encounter when there's no water flow carrying the heat away.