In the list of primary power sources in one of the previous entries, I dismissed temperature as a non-viable source except if the target device is placed in the vicinity of a regular source of high temperature (fireplace, furnace, etc.)

Turns out, I am wrong (as usually).

This clock is said to run without being manually wound up for more than one-and-a-half century, by now; and ticking. It uses an enclosed "bellows" (which could be filled by gas/air/vacuum/liquid+its vapours), which expands and collapses as temperature and atmospheric pressure changes. These movements then wind up the conventional weight-and-chain "accumulator".

However, this still does not mean this power source is practical. As noted already, the mechanical clock got optimized to minimal power consumption during the past centuries up to a very impressive point, which may be impossible to surmount by a practical electrical/electromechanical device. Also, the "commercially produced" version of this clock is a superluxury item, and this may indicate that  developing and constructing the "bellows" and associated mechanisms may be not entirely trivial...

This is an electrical device which runs uninterrupted (mostly) from 1840. It contains its own power source - a couple of high-voltage battery "pile" - and does not produce any useful work, but it does this for an impressively long time.

While this may appear to defy my assertion that devices based on electrochemistry can't last for a century. This again has to be viewed in context: that the "Oxford Bell" performed so well in long term is almost an accident, rather than result of directed research. The materials used - mainly the sulphur insulation - is not very viable for a practical battery. Modern batteries rely on polymers and other organic materials used as electrical insulation and sealants, and that's those materials which are the most likely to degrade/change in the long term. Also, the processes in primary cells are much more simpler and less prone to various side effects resulting in gradual degradation of performance, than accumulators and supercaps. Also, note, that these "piles" provide high voltage and miniscule current (and even that in short pulses and very low fill factor); while our electronics requires relatively low voltage and comparatively much higher and steady current. Also, we can't even copy the design, as we don't know for sure what's inside...

To develop a battery - yet even accumulator or supercap - guaranteed to last a century, would undoubtedly take decades of work. Given little to no demand, there is no way this will happen in any foreseeable future...