I wonder what the time course for cooling looks like. That could be pretty diagnostic. This group must have turned off the input power at some point and I would have expected then to see some interesting stuff.
Yes, that would be interesting.
Overall, this isn't what I would expect to see from a system with a source of excess-heat that is, itself, activated by increased in temperature. There should be an inflection point as the excess heat production starts to create a positive temperature feedback to engage even more excess heat. I see no a hint of anything like that.
What about the big event at around 5000? And what about the fluctuations? They are not seen in the calibrations. Perhaps they are mini-inflection points. The heat is turning on and off.