Potential energy includes dispersion correction term 2 times during minimization

Summary

I run energy minimization of the system using the l-bfgs algorithm with energy dumping every frame. I use long range dispersion corrections for energy and pressure. Then I use gmx energy to get all the energy terms. The sum of the energy terms (bond, angle, ...) doesn't match the full potential energy. It seems that the potential energy includes the dispersion correction term 2 times in the sum. During MD instead of minimization the potential energy matches with the sum of the terms.

GROMACS version

GROMACS version:    2022.4
Precision:          mixed

Steps to reproduce

gmx grompp -f min.mdp -c box.gro -p topol.top -o min.tpr -maxwarn 1
gmx mdrun -deffnm min  ; run 10-step minimization
gmx energy -f min.edr -o min  ; 1 2 3 4 5 6 7 8 9 10 11 0  to get all energy terms and potential energy

import numpy as np

log = np.loadtxt('./min.xvg', comments=['#', '@'])
energy1 = log[:, 1:-1].sum(axis=1)  # sum of all energy terms
energy2 = log[:, -1]  # potential energy from log

dE = energy2 - energy1  # difference (should be zero)
print(dE, log[:, 8], sep = '\n')  # print energy diff and Disp corr term (they match)

bug.zip

What is the current bug behavior?

The potential energy includes dispersion correction term 2 times during minimization.

What did you expect the correct behavior to be?

The potential energy includes dispersion correction term only 1 time.