From what ShyTorque and I have been saying it seems possible that the following applies:
We adjust the idle controls of mixture and throttle stop on the ground to give the desired RPM and smooth running as per the engine manuals.
We then go fly and change pressure, temperature, density and even vibration levels (poor old float wobbling about and so allowing different fuel levels in the carb) and g. The fact that the engine idles satisfactorily following such possible variations is perhaps due to the prop shoving it round and also providing a smoothing flywheel effect. If the prop for some reason is having its own airflow changes (big yaw or pitch effects like spinning) and is at very low airspeed anyhow then bingo the deficiencies of the current idle set up become apparent.