Thanks, davidfor!
You are correct, 1 and 2 pertain to the same basic problem and only describe two different ways in which the phenomenon occurs. I listed them separately because a software modification could help alleviate the situation: Larger touch zones for the locators would tremendously improve the highlight correction process, so this would be a first workaround until the touchscreen driver has been fixed.
The problem I am facing as far as continuous highlighting is concerned on the Aura One is that not lifting the finger is not enough. Often when I move the finger too fast, the process is interrupted because the touchscreen itself does not seem to be able to catch up. This has nothing to do with the speed of eink and just betrays a general laziness in the driver code. Kobo apparently did not deem it necessary to have the touch layer react much faster than the eink screen. Yet this would be an important first step to more reliable devices.
As to the design decision regarding highlights across pages: The Kindle recognized sentences and stops highlighting at the end of the first sentence and will not move any further until the finger has reached up and has come close to the highlight. Really worth checking out, by far the best implementation on any eink device. It automatically prevents the type of unwanted behavior that the Kobo shows without the need to improve performance all that much (although of course this has always been an issue with Kobo readers, they are just not as snappy and the Kindles).
I guess the first thing Kobo really needs to do is drastically improve touchscreen precision and speed. Of course, when I tell them that, they are working on it
(@FrustratedReader: The Aura One touchscreen as well as those on the Aura, Paperwhite, Voyage and Oasis are capacitive. Yet on the Kobos those are, in my experience, even less precise than the IR one on the H2O)