Temperature distribution after using GridFiller

[HomesGH]

Describe the bug
This bug might be related to issue #224.
When just setting up a domain "filled" with a grid, this could be unexpectedly difficult. See e.g. this example, where an "explosion" occurs in the first few simulation steps.
After having a closer look into this and coloring the particles depending on their temperature gives (Simstep = 1):

It seems like the particles at the edge are a bit hotter compared to the ones in the bulk. Furthermore, there are very hot (i.e. fast) particles in the corner.
I used MegaMol for the visualization and it should also consider the periodic boundary conditions when calculating the temperature. However, there could also be a bug in MegaMol.
This could be cross-checked with e.g. ParaView. However, it looks a bit better when using the old CubicGridGenerator (therefore, I guess that MegaMol is working correctly):

To Reproduce
Steps to reproduce the behavior:

1. Run the exemplary simulation
   
2. Observe the explosions and view the progress of the temperature during the simulation

Expected behavior
The equilibration of a liquid is very basic and should not lead to any explosions etc.
The state point of the exemplary simulation is more or less saturated liquid argon at a temperature of T=110K and a density of 31160 mol/m³. The saturated liquid density at T=110K is 31110 mol/m³.
This bug could also be an issue of reduced units or at least related to it.