Whoa, I almost thought you were gonna guess the answer! It had to do with temperature, definitely. I'll let you off the hook in a second- But first, here are some things I did to ascertain the problem.
1. I sat there for hours, and recorded the time that relevels occured, put 'em on a time-based chart, and found that relevels generally subsided logarithmically, meaning the incidence of relevels was high after a reset, but dropped off quickly, continued to decline at a diminishing rate, like an olympic ski jump. That rate of decline reminded me of temperature based issues.
2. But I still had the bubbles escaping from the jack packing! That loss of volume would cause it to relevel. Bubbles never subsided, they just kept coming "forever".

So where was the air coming from? After all, the pump is submersed in the oil, and the only suction is at the end of the pump, everything after that is pressure only. I theorized about a multi-sectional piston, with loose joints, where the oil is entering the piston cavity through a lower joint, and pushing air out of the upper joint. Sounds good, except this was a one-piece piston. Hmm.. I even looked at the pit pump return line into the tank, but that was way across on the other side, away from the pump inlet. Yes, on the other side, until the pump started! The suction and/or turbulance pulled the long return line hose all the way over to the other side, where it got sucked up to the pump inlet. Since this was a new job, there was no oil in that line, just air, and it sucked that air into the pump, blew it into the jack, filled the jack with bubbles, and the car would sink as the air escaped through the packing. Then the whole process would start over again, and it basically would just keep releveling until it faulted out. Cut hose shorter, no more bubbles. Thought I fixed it, but the exact same fault came back. Still releveling, still faulting out. Weird.. Conclusion tomorrow. I really gotta go.