I don't really see the big deal about this. Most voltage regulators can be made unstable by one means or another. A perfect op-amp and a perfect darlington might be unconditionally stable (in all circumstances). On the other hand, an imperfect op-amp and imperfect darlington might be just fine for such a device as the 7805 voltage regulator. Take a look at the innards of the uA7805: -
It looks complex but it has a darlington transistor output and a makeshift op-amp driving it. There is feedback to the inverting input of the makeshift op-amp and the non-inverting input receives a voltage from a reference circuit (far left).
The darlington is acting as a voltage follower, providing no gain at its emitter - why should this configuration be any less stable than an op-amp (unity gain configuration) with a load capacitor? Those types of circuits can be stable - it all depends on the op-amp of course.