OP said in post 1 that when EGC to pool subpnel light was lifted the 1 volt went away. He later said when disconnecting some of the other equipment bonds, the voltage raised to 3.5 volts.
Let ,e try to clarify that a little:
He said
If you disconnect the ground wire coming from house subpanel feeding pool subpanel
The 1V measured between the pool light and local earth went away.
Nothing about a pool subpanel light. I am assuming that the "120V light" is a pool light, not a light at the panel.
Anyway, in the absence of any other information, I would first consider that the EGC is effectively bonded to the light shell in the water, but that there is a tingling due to the voltage difference between the water and the earth around the pool.
Two ways to approach that:
1. Figure out why the EGC is 1V off local ground (there are any number of code compliant reasons for that given neutral current flowing in/out of the GES somewhere, not necessarily at the house in question. Fix that somehow.
2. Figure out why the local earth around the pool is not part of the equipotential grid connected to the EGC.
The second one is probably more valuable in the long run.
The information about 1V --> 3.5V with some combination of bond lifts is more unclear and possibly not relevant to this problem except to the extent that it affects point #2 above.
Does that fit with your understanding of the OP's posts?