Robin Leadbeater

Yes the spectrum will be a blend but the lensing object must be very much fainter (or spectroscopically similar) as there is no evidence of any change in the spectrum even though the total brightness (driven by a change in the brightness of the lensed component) changed by a factor of ~25  (3.5 mag)  I am not an expert on lensing but I think once we have a complete model we may have a distance and mass for the lensing component. That would also give an idea of how far back in time they would have been resolvable. Perhaps historical deep images would show something? (Of course not if it is a black hole though)