Well, I've learned a lot in the past couple of weeks playing with different parameters. The changes I made to the timing map has made the most overall difference. I started out just matching the timing between cylinders 1 and 2. I looked at which of the two cylinders had the most timing in any one area. In almost every rpm and throttle position that was cylinder two. Then I used the smooth feature, being careful that it did not create any large jumps in timing, especially down in the lower RPM and harder throttle positions. I also spent some time making the area where the engine idles a lot more stable, so that the timing isn't jumping around so much. The original timing map has the timing jumping as much as 10 degrees in as little as 200 RPM, as well as, as much as 10 degrees difference in timing between cylinders. My theory was that all that fluctuation in timing was contributing to the roughness and jumpiness of the lower rpm throttle response. Anyway, the bike is super responsive, smoother and hasn't stalled on me once since I made these changes to the timing.
The other thing I found, partly because I am doing all this in February here in Oregon, is that I was having trouble getting consistent results logging AFR and applying corrections in autotune, until I found and balanced out the "Coolant Temperature Fuel Compensation" in "Other Maps". The stock settings in there calls for a bunch of missmatching corrections for temperature, which was causing me to go back and forth on my autotune corrections depending on whether I was logging data in the 36F mornings or in the unseasonably warm 60F afternoons. After comparing the morning and afternoon logged ARFs, and analyzing the results, I tried balancing out the map. The idea is for the map to add fuel when the coolant temps are lower, but I don't think it should be adding a different amounts of fuel to the two cylinders, especially since I got rid of the disparities in the ignition timing between the cylinders. And since my results were actually showing that it was running quite rich when cold, I just matched the corrections on the two cylinders to the cylinder with the lower value. This has helped a lot when it comes to getting consistent results and the suggested AFR changes are getting pretty minor after just a few more logging sessions.