Does anyone have experience using bi-polar speakers?
Yes, I built 36 of them for my HW project. I'm glad I did, but I'm not sure the benefit would be as great in the smallish room you describe. For one thing, they can't be directly against the wall unless turned sideways.
While I don't know about the upgrade path of the Motu (from 24 to 48 etc), I remember reading somewhere that you could start with an AudioFire12 then add an AudioFire 8 or AudioFire4.
I strongly considered both of these before settling on the MOTU 24I/O. For the MOTU, the 424 PCI card supports up to four 24I/O units, so that option is expandable to 96 channels in 24-channel increments. I needed 48 channels, and in speaking with Echo's technical support, they couldn't guarantee me that four AudioFire 12 units would work, even if spread across multiple FireWire cards. At the time, they had only tested up to three AudioFire 12's together. You
can mix different AudioFire units on the same FireWire chain. The Echo has slightly better audio specs, while the MOTU supports cable lengths up to 60 feet. I've had the 24I/O's for over a year, and they have performed absolutely flawlessly on OS X.
In your case, I'd probably go with the Echo AudioFire 12 and some of the active monitors others have recommended here.