Dave, you're mentioning the possibilities of niche application such as remote sensing but I'm having trouble thinking of a realistic scenario where this would be sensible. For this to work, there needs to be a temperature differential, but unless there is a powered device that generates enough heat to produce significant temp dif involved, the temperatures in nature tend to equalize. Of course, if you have a powered device that generates tens of degrees of temperature differential, you might want to consider tapping into that power source. I'm sure that at these power levels, no one would even notice.
If you can't use a battery for whatever environmental of regulatory reason, wouldn't a solar cell such as the one in the calculator still be a more suitable solution? Charge a tiny supercapacitor for those cold nights or cloudy days.
Speaking of solar cells, energy harvesting and wearables, I have a Seiko solar powered wrist watch. The entire face is a solar panel and just by wearing it daily, exposing it to mostly indoor light, it provides enough energy to keep the watch running from early spring to late autumn (I need to remember to expose it to direct sunlight for a few hours in autumn to charge up the battery for winter). I'm fairly sure you've covered this in one of your previous videos, explaining that a simple and cheap battery is still more sensible approach and I agree, the watch is a bit gimicky, but it actually works. The watch itself has a microcontroller that regulates movement, so when the battery is depleted, it first starts moving the second hand two positions, that waiting for two seconds. If you still don't recharge the battery it parks the second hand a 3 o clock, but it still counts the time for a while so once you recharge the battery it moves the hands to show correct time. So it's a fairly complex electromechanical device. The solar panel is about 27 mm in diameter, so about 5,7 cm^2.