Partial answer.
If we look at both existing, Earthbound optical and radio interferometers together, we see a general trend. When all the "collectors" (telescopes or dishes) are within of order one hundred meters or less from each other, optical beam lines or radio transmission lines have been successfully used to combine the signals.
For longer baseline interferometry, the signals are usually down-converted and digitized locally (say a few Gbit/sec bandwidth max) before they are transmitted to a central location. ALMA and the future SKA use fiber optic cables.
For longer baselines like Event Horizons Telescope the data is locally recorded on old-fashioned hard drives, then flown to a central location where interferometry is implemented off-line in a big computer. This required very careful determination and recording of the relative positions of all the dishes around the planet.
An exception is Spekt-R which down-converted, digitized, then live-transmitted the 2 x 72 Mb/sec data to Earth over the sometimes very long baseline, only while it was visible to the receiving ground station.
As I understand it we can combine the image from multiple telescopes into one, increasing the resolution, but not the brightness. The further away these two telescopes are from each other the better resolution, and as I understand it we've done this with Earth based telescopes.
Yes that's right. I don't yet know how optical imaging data is combined interferometrically...
but in optical interferometer "beam lines" without digitzation, I believe you need periodic placement of relay lenses because light diverges whether you want it to or not. In space over long distances, conservation of etendue prevents us from optically trasporting an image over a very long distance.
That means for optical as well as radio, you need to down-convert the eletromagnetic radiation to a manageable frequency where it can then modulate (either digitally or via analog techniques) a stable laser beam (analagous to the fiber optic lasers used in ALMA), then shoot that free-space optical connection between your satellites.
You might get a THz of bandwidth, which would correspond to say a 1 nm wide band at 1 micron, so you won't be doing broadband imaging.
In my personal opinion if you had several billion dollars in cash handy, you could put a few satellites in orbit based on existing technology. It would take time to get it to work well (the devil is always in the details) but there is no fundamental show-stoppers here beyond sufficient funding and brainpower.
