OK, job completed. When I opened the light fixture there were two bulbs out. One was dead, the other needed slight tightening to come on. So the flash could have been the first one burning out. Since one bulb was still on it appeared the light was normal afterwards. Or it could have been that the loose one had an intermittent contact, but I think the first idea is probable.
So it seems there is aleays an unexpected answer to a strange symptom.
As to the EMF part of the job, there were net current fields from a number of linear sources: circuits as well as pipes.
One error: even though the main disconnect was outside at the meter, and the ground rod GEC connected there, the electrician had added a GEC to the inside panel neutral bus.
This clamped onto a water pipe and then ran back to the outside box's ground. So this created a parallel path for neutral current from panel to meter box. Also a green ground run from panel neutral to the meter box, also carrying neutral.
All these neutral paths (pipes touching each other) were eliminated by disconnecting the GEC from the neutral bus.
The other net current path, which did a loop around the building in the basement ceiling, was all due to having wire-nutted a neutral from another circuit to that light circuit (with the flash), thus bleeding off some of the neutral current to the other circuit. Solution: disconnect that neutral connection.
Karl
