The short answer is that CO2 leaks out of tubes significantly faster than compressed air. I have personally tested this under somewhat controlled conditions and proven beyond a reasonable doubt that this is so.
Why this is the case, I cannot say. But good data beats a good theory every time.