1. My 100L seems to need more afma the longer the lens-subject distance gets. Did I do something wrong, is this normal behavior so I actually arbitrarily need to pick my favorite distance I'd like the lens to af? Do I need to test my 70-300L at longer distances than the 2m right now for accurate results?
It's normal for AFMA to vary with distance; at longer subject distances, the delta becomes very small, which is why the usual recommendation is to test at 25x to 50x the focal length. But if you almost always use a lens at a given distance, test at that distance. FWIW, my 100L macro on my 1D X needs +4 at 25x, +2 at 50x, and +3 near the MFD (so I use +3).
25-50x the FL means you should be testing a 300mm lens at 7.5-15 m, not 2 m...
2. I don't understand the scale of the afma values - does for example make the +5 of my 100L @1m a difference at the dof of f2.8 in real life? Or is this overcompensated by the camera's and my human's focusing variance anyway unless I shoot on tripod (and in this case would use focus peaking or contrast af)?
The scale for AFMA is one unit = 1/8th
the depth of focus for a given lens at max aperture. Depth of focus is similar to depth of field, but the former is measured at the image plane. In practice, I'd expect to see a difference with a +5 adjustment for an f/2.8 lens (but it would be barely perceptable for an f/5.6 lens, if at all). In part, that's due to the precision of the AF system itself - a high-precision f/2.8 AF point is specified to be precise within 1/3 the depth of focus, while a standard precision f/5.6 AF point is precise within one depth of focus.