It was done with one single dish,
...in Ku-band (16.7 GHz) and ... currently equipped with a high target resolution.
The Fraunhofer Institute for High Frequency Physics and Radar Techniques (FHR)'s March 21, 2018 press release - the original source of the picture in the question - links to an introductory description of their Space observation radar TIRA ...
Full disclosure: ICEYE employee, I'll try my best to answer. Thank you for asking and being interested in ICEYE-X1!
The "amount of seconds" refers to the amount of time the satellite is actively imaging data for an individual frame.
I would not recommend referring to the current Spaceflight101.com article for technical specifications on ICEYE-X1.
ICEYE-X1 does not have a vertically large antenna, so it can only provide 2D data. To interpret the image correctly, you can imagine the satellite crossing over the top of the image from left to right, looking sideways. The axis are:
x-axis: position along track
y-axis: slant range
color axis: reflectivity
None of these measurements requires a large ...
A few issues with what you included:
The resolution is stated as 10m in the bit you included.
The size of the image was 80 km/ 40 km
The frequency is 10 GHz, meaning a wavelength of about 3 cm.
The SAR antenna is deployable, meaning the effective beam width can be greater then the satellite. Of course, the total width is about 3.2 m, as you indicated.
You could use a main satellite with 2 subsidiaries. The subsidiaries use orbits with the same shape as the main orbit, but in a plane that has a slight angle relative to the main (plus a minimal difference in altitude to prevent collisions). The point where sub 1 crosses the main's orbit should be 90 degrees away from where sub 2 crosses the main's orbit.
If you are looking perpendicular to the orbital plane, instead of radially outward, you can have your secondary sensor at a lower orbit by your chosen distance d. It will drift ahead of the craft by about 3 pi d for every orbit. For LEO it only moves ahead by about a tenth of a second per orbit. So the images from the secondary sensor slightly earlier so ...