Hi,
I think it is possible to build a telescope that can distinguish light sources and their properties based solely on statistics and correlating basic properties of the detected light over a period of time. For example if the light detector can do a basic distinguishing process for the incoming light then over time it can average out this to build up a higher resolution of that property.
The simplest case would be a 1 pixel wideband light intensity detector, over time as this is pointed at the sky, it's output will be proportional to the sky light hitting it. If run like this for years, with a lock in amplifier to reduce noise, over time it may be possible to detect the position of stars based on their minute (below the normal noise threshold) impact on the detected light as they cross from one horizon to the other. All of this data would be correlated and result in the signal that a star exists at a given coordinate.
To take it a step further a polarization sensitive detector could be used that can also distinguish sources based on polarization or other properties of the light.
I think this type of "telescope" is already used in particle accelerator detectors, but the detectors are receiving a much higher rate of data and thus can correlate it much faster, whereas an optical telescope aimed at the sky like this may take decades to be able to build a rudimentary star-map, however it has the potential to theoretically detect all light sources given enough time I think.
cheers, Jamie