I don't know if it would be a problem in practice, but I'd be concerned about how high a frequency you need to spin the motor at for the imbalance to be detectable, and the resulting frequency at which you need to read the accelerometer.
I wonder whether you could use some sort of piezo mounting as an AC signal generator and then use a peak detection circuit to trigger a strobe when the peak occurs. What that doesn't do is give you any idea of the magnitude of the imbalance, which of course you need to know to make any correction and also to know when the balance is good enough. Maybe you could detect the signal level as well? Quite a lot of electrical hardware to design though, to get it working.