Yeah, we normally do (at our club) 4 rolls with 300 ml, or 6 rolls with 500 ml (that may involve even four to five different processing sessions; between them, developer is poured back to the bottle and reused later). One "roll" means 135-36 or 120. So 220 is "two rolls".
This is just like Tetenal C-41 specifications as well as Fuji C-41 specifications say; you can process about 12 - 16 rolls per liter. We have had no problems, but, as I said, no side-to-side comparison done. We have been quite careful about chemical contamination and washed the Jobo processor well. We use a manual processor Jobo CPP-2. The nice thing is that when you do the final rinse/stabilizer outside the processor, the final wash will also wash the processor quite well.
Fuji's instruction says that one-shot approach should be used when developing "professional films" to "professional standards". I think that this is because the time compensation doesn't necessarily compensate all the color layers similarly (look at http://sorsa-tv.ath.cx/~antalh/siwassakayntiaika.gif ; this is for ECN-2 film but you get the point); and it's hard to know just how much to compensate (it depends on the overall density of the previous films). However, I think that these errors are so small (crossover maybe about 0.01-0.05, density maybe about 0.1) that I reuse my developer. I can't see any difference. I guess that developing film in a random minilab will introduce much, much larger deviations from the standard.