I agree with Andy, seems pointless to do that since it will learn under real-world driving conditions. Furthermore, those 'real world' driving scenarios would likely not be the same as a dyno cell so you'd be muddying the waters before you even put the thing in a car. An engine dyno can't simulate differences in driveline frictional losses, weather, car weight, aerodynamics etc. A chassis dyno might be better suited to that type of testing if you were trying to establish a specific fuel map since you can simulate load better. But like he said, if you have a known 'good' curve' to plug in, it accomplishes the same thing (establishes a baseline) and the auto learn would adjust appropriately as you drive it.