I also had to reduce the tested set of target_brightness values by 1 each, because the new API as it currently stands will range not from 1 to 255 for the mocked display like the previous API, but from 0 to 254 with the minimum already being incorporated into (i.e. subtracted from) the brightness value that the interface deals with.
The tests still fail locally for me because of memory leak errors (mainly in Python and ddcutils, apparently) but CI now seems to pass. Yay! Should we squash the changes into the main commit series?