About 8kHz sampling.. If i remember correctly, it was default rate for ac and generated spikes to logs at random. Might happen every minute, or might go for hours between noticeable errors (at 10sec intervals, probably appears hell of a lot more at max rate). And it (mostly) affected both channels at the same time (and in my case, always in different directions), so it was “easy” to spot from graphs. If you are logging only one channel, set other one to internal temp, so you have something to compare, where every spike has to be an error.
So it is possible to use 8kHz, if you know what you are measuring and know how to filter those errors out. I believe that is the reason James still left it as selectable option, if someone wants that precision and is willing to double check their results..
And when trying to max out mooshimeters measuring rates, its good to remember it was never told to be an oscilloscope ;)