“…because of that tenuous atmosphere she reaches the required speed much faster, due to lower drag. I'm still not convinced aerodynamic control is impossible, but I'll let the boffins give the final word on that matter. I'm a mechanical engineer, not an aerodynamicist.”
As someone with (a little) experience launching rocket planes, I will tell you right now that even at or below sea level, aerodynamic drag would never be the limiting factor — simple dynamic equilibrium, Newton’s Third Law, is the problem.
Look at the chart at http://0.u.is/_7p42 — the curve of interest is the one labeled G12-RCT. (They have said they are using a more powerful motor, but let’s use this for now). The initial thrust, right after ignition, ramps up to max power in a couple of milliseconds (see how the curve appears to intersect the vertical axis on the graph) so we can consider the thrust to be 8 lbs, roughly 40 newtons.
Given an aircraft weight of only 500 grams, which will be MUCH less than it really is, how fast could it get over three metres’ acceleration if we ignore all drag sources? Well, 40N/0.5kg = 80 m·s^-2 or about 8g; this sounds impressive (and it is), but let’s see how fast it’s going after three metres’ acceleration at 80 m·s^2:
x = v0*t + .5*a*t^2 but v0 = 0, therefore t = sqrt(2*x/a) = sqrt(2*3/80) = 0.274 seconds
Its speed will be 80*.274 = 21.9 m/s = 78 km/h or about 50 mph.
Can we agree that this is basically the minimum possible speed we can expect aerostabilization — that it would be silly to expect a rocket-plane in 5% of an atmosphere to be stable at under 80 km/h? Then it means we have to keep the mass of the plane below (500g/80N)*[initial thrust of final chosen engine] for it to work. And I don’t think that’s feasible.
Now what happens if we put in a piston launcher? The initial thrust gets immensely magnified until it’s essentially an impulse of huge magnitude, 10 times greater at least. With my small model (and a ground-based launcher) I got the plane up to ~60 km/h in about 30cm. The benefit of an air-launched one would diminish since the pad would launch itself backward — it would depend on the mass ratios of the two parts.