Actual over-the-horizon radar systems have used transmitter and receivers at a significant distance from each other -- the Soviet Duga radar system in the 70s had its transmitter and receiver separated by ~50 kilometres or so. The receivers (and in active systems, also the transmitter) need to be in very precise time sync, and being physically close enough to run a direct cable makes that a lot easier.
But theoretically, assuming the system is sensitive enough and time-synced enough and you have enough computational oomph available, an arbitrary number of receivers, at any distance, can be used to synthesize an arbitrarily large aperture. It's actually a similar matter to the synthetic aperture radio telescopes used in space astronomy, where telescopes with an effective aperture ~100 million kilometres wide have been created, using space-based radio telescopes linked with ground-based radio observatories.
Such systems may not have the sensitivity of a 100 million kilometre wide telescope mirror (no matter to catch the photons since the "telescope" is mostly empty space) but it does have the equivalent angular resolution of such a 100 million km wide telescope mirror. It's mind-boggling when you consider the consequences of this as computation power improves. It will soon be possible to do this at such high speeds that it will allow frequencies into the far-infrared spectrum, not just microwave, for example.