I'm not an expert, but here is what I think I know.

The shorter the wavelength the harder it is to measure phase attributes. Hence, optical interferometry lags radio interferometry. Current optical interferometry plans revolve around measurement difficulties. Nulling (to get rid of central star light (that whites out planets) and recombining the beams physically is currently the most effective way to do the optical interferometry targeted at planet imaging. If the measurements could be done accurately enough in place then what you suggest would probably be possible. One caveat might be the amount of data involved might swamp even our current computer storage capabilities especially factoring in having to transmit all this data from space. Obviously, the collecting area affects the quality of the picture and the practical ability to build up a picture. The more photons the better in terms of quality and so there are some practical considerations in terms of the size of the collecting area with regards to the faintness and distance of the source.

For the foreseeable future there will be place for analog or physical designs to effectively filter the enormous amount of data in the physical world. Doing the recombining physically is in effect processing the data like a quantum computer--the amount of computation saved on our current technology would I think be enormous. If you had perfect detectors that could detect every wavelength and attribute of every particle hitting it, then theoretically all it would take is (potentially enormous) computation to become any kind of instrument. Maybe someday we will have atomic scale sensors and quantum computers and will have such generic instruments. But that is a long ways off.