I think I found it yesterday, but not sure yet till weather and workload cooperate for testing.
Looked at mag and had a lot of oil in there. For last 2 or 3 years, factories have been drilling the mag mount holes right through into the inside of the case, so that the flat side of the armature is all that seals those holes. In cases where the armature is a bit crooked and hits the case at other points, I've had to put 4 thin washers between armature and case so that all 4 screws will seal.
I pulled mag off & fitted a new one and think that may work now, but didn't look closely at old mag. Later, I noticed that the 4 rivets had never been seated into the armature, leaving needle sharp points on the case side and smooth rods out the front. There were some burn marks around the rivet that ground was soldered to and around some of the points. I'm thinking the return path of spark was arcing there and got hot enough for the ground to fail.
I put about 8 or 10 tons on those rivets and resoldered ground wire. Now all is straight and flat, so I'm thinking it will be OK when I get it back into the bike.
The mystery came from the fact that the meter didn't push enough voltage to cause the arc, so always read just right when testing. I'll post again if it fails.