So, I replicated your test conditions and got pretty much the same thing, only worse.
I was using a ~ .07 Hz sine and found "The jump" occurred at around + 300 uV on the leading edge and + 100 uV on the trailing edge.
About 50 % of the time I would get a big jump (~100 uv and then it would fall back , in other words, go in the wrong direction.
I used the Uni-T utility just to make sure I wasn't seeing any bugs in my ongoing software development.
It strongly looks like some sort of hysteresis, but it's tough to tell if it's software or hardware. The anomaly doesn't occur at zero, nor is it symmetrical about zero, so if I had to guess, it's something to do with the digital calibration removing the DC offset, fighting with some hysteresis in the in the front end, but that's just a SWAG. Because it doesn't happen all the time, it seems to be timing interaction between when the input crosses zero and the meter takes a sample/reading, likely a settling time issue. It doesn't seem to get "stuck". which would be a real problem.
When measuring a similar DC voltage, it is tough to tell how fast it responds to slight changes. At 800 uV, you're kinda in the weeds and I don't have a fully shielded set-up for this quick test. That said, it does appear to stabilize fairly quickly. When I have time, I will test for monotonicity or missing codes. For now I'm a little disappointed, but I think it's just something to be aware of.