I tried an experiment. I set my idle coast table up so the values in the cells where the spikes occured were now 1 degree higher than the High Octane table values. What I now see is 1 degree upward spikes when the surge occurs. So I'm confident that the changes in timing seen in the logs are a result of the car transitioning from the High Octane Spark table to the Idle Coast Down Table. Not a eureka finding and it doesn't explain the surge but another data point.
Why it does that....dunno
For DSTECK: How low did you go with your timing in the areas affected by surge?