Armxnian, while it's true what you're saying...those measurement methods could be done, I don't really see the point. In the end, you're trying to find a settled value and the integration time has a direct correlation to the settled value of a specified period of time. Siglent says they do 5 sample per sec on slow mode, but they only update the display about once a sec. So there is clearly a mismatch. If this is true, they are either averaging about 5 reading to get 1 update per display, or they're doing as you mention and only showing one value of the 5...which means they cannot resolve more than about .2s.
On other topic from Martin's video. I don't have the voltage settling issue he notes. I added a very large RC on the output of my supply and let it settle for a long time, then hooked up the SDM3055...it goes almost directly to the final result, no settling. There does seem to be bit of a jump in DCV when the filter is applied, but it doesn't jump much.
In looking at some other things, I used the Easy DMM software to capture some data. It doesn't seem to matter what capture rate I use, it's always about 50 samples per second. Not sure what that's about, but the display on the unit and the display on the Easy DMM shows pretty much the same thing. Strangly there is a weird bobble when the system goes through a zero crossing in slow mode, on both sides of zero. This can be rectified by going to either middle or fast, or raising the DC offset of the system. See the attached picture.