I think the three of us are on the same page now - what a glorious day! You don't want to mess with electricity in the dark? LOL, just teasing. ... "Post #140" - more LOL - I never even realized they are numbered! .... Anyway, while we await the ladder-only test result (only the ladder bonded back to the pump, no wire laying in the pool), let me please make a slight modification to my answer about whether "the light could still be the entire issue". Same answer with regard to bonding, as I think we proved the ladder is not bonded, and possibly the rails too, the light still unsure. But early on we were also chasing the original source of the NEV, concluding at this time, I believe, to be the power company, and which is nearly unavoidable, variable, etc. So we appropriately switched to chasing the bonding issue which is far more important anyway. But it seems to me now that any power co NEV could be propagating through the ground wire to the light too, similar to the way it propagated to/through the pump. There are a few problems with how that light is wired, and I propose a couple ways to test.
The pictured junction box (post 145) is not listed for pool use, it's just a vanilla exterior wet location (metal) box. Most electricians interpret NEC 680 23 & 24 to require junction boxes that are listed for pool use, just like with transformers and other devices. Somehow the installer perhaps thought otherwise and the inspector missed it or interpreted code differently. Regardless of the box used, a wire nut to join the bonding wire with the other ground wires is not sufficient, they are to be tied on a small buss bar, and in the case of a metal box, the buss bar is to be integral to the box - all so that all those connections, including the box are secure. Samples pictured below. Next, when a #8 bonding wire travels through a conduit (with other wires) the #8 is to be insulated (green) - it appears they used the bare #8. I'm not sure how important all that is, but I mention it all because if testing shows the light is also not bonded, a new and proper box replacement could help solve the problem. On to tests of the light bonding....
I would test for light bonding by having all those J-box connections loose, disconnected, and measure resistance in ohms at two points. First, with a probe tied tightly to the loose Jbox bond wire, place other probe against light ring (or screw for the light ring - yes you're in the water) - it should be near zero ohms, and probably is. The more important test is to measure ohms from the loose #8 to a known bonded item such as at the pump (or to your temporary "array-o-bond"). With good probe connections this too should be near zero, anything more than an ohm or two probably means the niche itself was never tied in (ie bonded) underground. When making that test there may be another path, eg through the water but that path would be rather high resistance, perhaps between 20K-100Kohms or more and same conclusion that they did not bond the light or the bond deteriorated. Fortunately the remedy could be easy if you have the box replaced, as the lug on the bottom of the box could be used to run/bury it around the perimeter (and pick up the new ladder bond (and rail bonds if needed) along the way. You could also test that now by tying the loose #8 to your temporary bonding array, then retest for voltages. If you pick up one of those little buss bars it might make some of your test connections easier.
I don't have much insight about the coping, but in the Mike Holt video I know the deck and surrounds like that got variable results, changing also as they splashed some water onto the deck. I suppose it might be tempting, while cutting cement for a ladder and other bonds as needed, to tie in the coping somehow?
