I wouldn't want to use ordinary chemical thrusters for several reasons. First, the thruster spew would probably contaiminate the optics. That's why the Hubble has no propulsion system. Second, even tiny microthrusters (milliNewton range) would likely have to fire a lot to keep the spacecraft in the proper position. Over time, that uses a fair amount of fuel.

If they intend to use thrusters to maintain super accurate positions, perhaps some form of electric thruster would be applicable, say ion or Hall effect. These have the advantages of very high Isp and very low thrust.

On the other hand, perhaps there's another way. Instead of trying to maintain micron precision with respect to one another, maybe it'd be possible to compensate for small positional errors using computer processing and/or adaptive optics. The thrusters would be used to maintain position within a certain range and then then the corrective systems take care of the rest. I'll have to read more on optical interferometry to see if this is possible.