Maybe for course corrections where you need to fine tune the orbit a couple of mm/s here and there?
For unmanned missions it does make good sense as the mass doesn't grow so strongly if the trip time gets longer.
Thinking about that I think I figured it out. The earth is going 30 km/sec around the sun, so that's probably where the 30 km/sec comes from. 30 km/sec is well past single stage to orbit and they would not need a booster if they could get into that range.
Mars is doing 50km/sec around the sun, but you can cheat in a few places as your earth orbit can add ~8km /sec from orbiting around earth. That's 50 - 30 - 8 ~= 12km/s plus delta V get out of suns and earths gravity well. But you can also play with orbital mechanics to save some energy and you can also subtract the orbital velocity around mars.
Anyway, I am not sure how much delta V you need, but it's well under 30km/sec.
PS: Video had refueling in what looked like LEO, but if your in a higher orbit the craft needs less fuel to get to Mars from that orbit.
The other option, is to have a pure transit vehicle from earth orbit to mars orbit and then a 3rd vehicle to land with. This, would work really well if we wanted to really colonize mars. But, if your only sending say 10 people the R&D is not worth the fuel savings.
PS: Let's say it's a 20,000 lbs craft or ~1/10th of the shuttle and your getting ~8.8 N or 1 lbs of thrust for 120 days. That's 9.8 m/s / 20,000 * 60 * 60 * 24 * 120 = 5k m/s. If you already need the panels while in mars orbit, that seems like a no brainier. Finally, If that's 200kw in mars orbit then the numbers seem even more reasonable as that's 460 kw near earth and the extra 260kw ~= 26 newtons which could add ~5km/s to a 60,000 lb spacecraft.