Nice report. I have experienced similar results using the Berger bullet and method.
I found the groups would start out diverged when on the lands or -.010 in. off.
I just kept on going until I found the best group. I would see my groups converge and then begin diverging again.
I would take the best group and split the difference between the preceding OAL and the succeeding OAL and shoot those OALs for groups. What ever the distance was from the two OAL, I would divide that by 6 so I had three groups below the best one and three above to see if there was improvement on either side of the best one determined by the first Berger suggestion.
After finding the best one of those, it was time to begin tweaking powder charges.
Pretty fascinating stuff.