FWIW, I tried having a red and green board on the same TB, and couldn't tell. I didn't test all that deeply, but in playing marble madness it seemed okay. I tried both (green,red) and (red,green) for (X,Y).
You are right there are performance differences, though. According to Happs, the reds have a higher backspin speed than the greens, and filter noise better. But remember you're hooking the TB to the computer through a hacked mouse (I tested optipac and happs usb interface but same idea), which has it's own backspin level, and filters the noise (as long as it's not confused by the noise in the first place).
The only things you might see (if you don't change the encoder wheels):
- small delay difference (in microseconds)
- having "one pixel" back & forth flipping or not
- max speed difference (depending on your mouse's max & backspin speeds)
- backspin speed difference (depending on your mouse's max & backspin speeds)
In my testing, I think the last two were not noticable to my interfaces (IE: probably my interfaces' max/backspin levels were lower than both boards'), and the first two were not noticable to me, even it they happened or not. You might get different results on your hacked mouse, though I doubt it.
My 2 cents.