My guess is that we've gotten out of the era where automatics get inherently poorer mileage just by their very design. As much as I hate to admit it, modern auto transmissions are a lot more efficient than they used to be. Audi/VW's DSG is even better, but that's not the issue with the A5 obviously.
My guess is that the manual and automatic transmissions in the A5 are very commparable in terms of mechanical efficiency. Which means that the mpg differences would be due solely to driving habits instead of mechanical concerns. The modern method sof EPA testing are supposed to reflect real-world behaviors better than the overly-controlled tests of the past...this amplifies the effect of any differences between how people drive manuals and automatics.
And when you get down to it, the car is smarter (and more consistent) about shifting for improved economy than the average human. I would concede that an automatic transmission selects shift points that are more optimum for fuel economy than I typically do. I usually run it up the tach more than I need to, especially in a car with a lot of available power. That's because the more power an engine makes, the earlier you can shift, since the car has enough power near idle to move it effectively. Shifting any later is wasteful...and it's easy, even natural, to shift "too late" in a powerful car.
Bottom line is that the apparent difference is probably due to humans being less perfect than computers. If a person drove the manual A5 with the express intent of maximizing fuel economy, the numbers would probably be equal to or better than the automatic. But the current EPA test procedures aren't designed to be optimized, they're intended to be realistic.
Then again, it could be something as simple as a different final drive ratio. But I'd expect that to affect the highway number as much or more than the city number.