@Deathrey
Quick answer:
I am not sure what is your expected appearance but heightfog one must be wrong anyway since it is done with simple height distribution and some fake lobe for the directional component and it does not feature multiple scattering. All these techniques (height fog, volumetric fog and SkyAtmospehre) have different ways to represent participating media, do the computation and different features (phase function lobes, altitude distribution, multiple-scattering) so brightness will change across them. Heightfog is a different participating media than the SkyAtmosphere on for instance, and it is flat not round, etc.
HeightFog and VolumetricFog should however get closer because they represent the same flat participating media (although the volumetric fog can represent more from voxelised entities) but they still can look vastly different because a different directional lob (or phase function). Maybe something else is missing also, we are actively trying to test more interactions by adding more test to our test suite also (the interactions matrix is huge).
So that 6.28 is a setup from the flat-earther!