Having performed the best optical lens focus I can (Adjustment-Service-Installation-Projector Distance-Optical Lens Focusing), I followed (I think) Curt's advice and did the electromagnetic focusing (Adjustment-Service-Random Access-Selected Source-Focusing). Pushing the adjustment all the way to one end (i.e. 99 on the adjustment scale) gave the best result, which is still not good.
So, from what I can tell the text in Adjustment-Service-Installation appears focused on the crt (looking inside the lenses) and at the same time out of focus on the screen, so that I cannot correct turning the lens barrel (screen photos are the screen projection of the crt photos). Electromagnetic focusing (Adjustment-Random Access-Selected Source-Focusing) improves on focus on the screen, but still not good.
Curt, do these observations shed any light / change your recommendation?
Looking into the green CRT (differences due to handheld camera angle to phosphor)
Looking into the green CRT after Adjustment-Service-Installation-Projector Distance-Optical Lens Focusing-Green
Above image as projected on screen at best focus with lens barrel rotation
Following above, best electromagnetic focus (Adjustment-Service-Random Access-Selected Source-Focusing-Green-Midpoint)
For the record, source is the HD4000 in Ivy Bridge (1080@60p) - HDMI - HDfury3 RGBHV - D15to5BNC - source4. For some reason (before any attempt to readjust the pj after relocation) when HDfury3 was set up to output component, there was a distortion horizontal band just below the middle of the image, so I switched the HDfury3 to RGBHV. While the HDfury3 was outputting component (YPbPr), I tried having the yellow and black BNCs also connected, but I got no image (although my understanding was that in component mode there was no signal in those cables and should function whether left connected or not).