It's not close enough to just put a sensor out in front, but it's definitely useful. Lots of multi-mirror telescopes use several conic sections; if you needed maximum light you could probably build a telescope with a monstrous 10km ellipsoidal primary, and let that bring the light down to a <1km secondary, and then bring the light back to an aspheric or hyperbolic tertiary optical system built with conventional mechanisms to actually focus the light.
However, I don't think you need to cook your optics with that many square kilometers of background radiation if you don't mind a longer exposure. Instead, you'd be better off installing a few small subsets of that hypothetical single mirror 10s of km away from each other, tracking the distance of each satellite through interferometry, and then putting it back together in software.
the data volumes are probably too high to make this work well. interferometry requires ridiculous amounts of data (hundreds of terabytes per image) and doing that processing in space will be tough within the thermal requirements.
