Number of degrees of freedo too small for 1D and 2D systems
With center-of-mass-motion removal (on by default), grompp will subtract DOFs for dimensions that are completely frozen. This leads to a too small, and zero for 2 atoms in 1D, number of degrees of freedom. This in turn leads to incorrect temperature coupling with MD. SD and BD give correct sampling, but might report incorrect temperatures.
Correct nrdf for 1D/2D systems
With COMM removal, grompp would subtract degrees of freedom also for
VCM groups with fully frozen dimensions, i.e. 1D/2D systems.
Also fixed division by zero for groups with #DOF=0 with VV.