Looks like the issue is high humidity and a defrost cycle that is too short, causing a buildup of frost/ice on the evaporator over time. My thought is that the thermodisc could easily be replaced by a permanent jumper (like cutting the original termodisc off and soldering the ends together) thereby causing the defrost cycle to run for the full 35 minutes instead of terminating when the thermodisc opens at an arbitrary temp.
I say arbitrary because my test of a replacement unit indicates the open/close circuit does not reflect the parameters stenciled on the unit. The stated specs are 140 & 30, but my test indicates 108 & 65. It does not appear to be possible to accurately determine the cause and effect in this case.
Very curious about this part as a replacement for the original part which has a spec of 65 & 20. Some engineer must have determined that the 65 & 20 spec was, at least in part, the cause of the poor defrost capabilities of these units. BTW, the part fits a lot of GE boxes.
It can take weeks for the issue to reappear; I will try to post results in a few months.