One potential problem here is that the raw analog readings may be noisy; although the user code can filter the raw values to remove that noise, this won't affect the changed flag, which is then used to wake up OLEDs. Not sure how to implement the support for such filtering here while also keeping the common case simple; ideally potentiometer_update_user() should be able to decide whether the analog input activity is actually significant, but there is no good way to return that data from there without adding some pointer argument.

(In the most generic case the user filtering code may also want to see all readings, including the ones which were completely identical to the previous value, so that it can distinguish that case from a delay between readings.)

Filtering was something I was planning to add, and the implementation there is mostly just a stub till a few other things are decided.

The main one being if this would be better as just a high level api to the adc driver. The assumptions that can be made, especially user facing ones, are somewhat simpler if its just for pots.