Error: Failed despite same graph (STOP 4) in SMEFTsim for operator involving gluon fields

Dear MadGraph authors,

I am trying to generate interference and pure quadratic samples for the operator cHG defined in SMEFTsim topU3l UFO model [1] and process:

import model SMEFTsim_topU3l_MwScheme_UFO_b_massless-cHG_massless
define p = g u d d~ b b~
define j = g u d d~ b b~
generate p p > e+ ve j j j j SMHLOOP=0 QCD=2 NP=1
output emVjj_cHG_SM_LI_QU

Although we are aware of the problems related to these kind of operators involving gluon fields but without dependence on as [3], it is still not clear the source of the following error:

Added good helicity 208 89.392844928797473 in event 1 local: 1
RESET CUMULATIVE VARIABLE
RESET CUMULATIVE VARIABLE
Error: Failed despite same graph: 1296
Have jets (>0) 0 0 0 0 0 1 0 1
Should be 6 7 8
STOP 4

This happens for MG5 version 2.6.5, 2.7.3 and LS 2.9.11.

The puzzling fact is that the MG reweighting at LO works, starting from the SM and reweighting to cHG=+1, cHG=-1 [4].

We tried the following combination for initial and final state partons but we can't find a clear connection with the success or fail for the execution of MG5:

define p = g; define j = g u u~ d -> OK
define p = g u; define j = g u u~ d -> OK
define p = g u d; define j = g u u~ d -> OK
define p = g u u~; define j = g u u~ d -> OK
define p = g u u~ d; define j = g u u~ d -> OK
define p = g u d u~ d~; define j = g u u~ d -> ERROR

Could you tell us what is happening behind the curtains? Should we "thrust" the result from the LO reweighting?

Thank you,
Best

--------------------------
[1]

https://gboldrin.web.cern.ch/gboldrin/generators/SMEFTsim_topU3l_MwScheme_UFO_b_massless.tar.gz

[2]
https://answers.launchpad.net/mg5amcnlo/+question/690133
https://answers.launchpad.net/mg5amcnlo/+question/687936

[3]

Starting point:
import model SMEFTsim_topU3l_MwScheme_UFO_b_massless-SMlimit_massless
define p = g u u~ d d~ s s~ c c~ b b~
define j = p
generate p p > e- ve~ j j j j SMHLOOP=0 QCD=2

We change process and model at reweighting:

INFO: change model SMEFTsim_topU3l_MwScheme_UFO_b_massless-cHG_massless
KEEP: change process p p > e- ve~ j j j j NP=1 QCD=2 SMHLOOP=0
INFO: Event nb 0 current time: 03h12
INFO: Event nb 10 0.16s
INFO: Event nb 20 0.28s
INFO: Event nb 30 0.37s
INFO: Event nb 40 0.41s
INFO: Event nb 50 0.45s
INFO: Event nb 60 0.52s
INFO: Event nb 70 0.56s
INFO: Event nb 80 0.62s
INFO: Event nb 90 0.68s
INFO: Event nb 100 0.74s
INFO: Event nb 200 1.3s
INFO: Event nb 300 1.8s
INFO: Event nb 400 2.3s
INFO: Event nb 500 2.8s
INFO: Event nb 600 3.3s
INFO: Event nb 700 3.7s
INFO: Event nb 800 4.1s
INFO: Event nb 900 4.4s
INFO: All event done (nb_event: 1000) 4.7s
INFO: Event /srv/emVjj_ewk_qcd_dim6/emVjj_ewk_qcd_dim6_gridpack/work/process/madevent/Events/pilotrun/unweighted_events.lhe have now the additional weight
INFO: new cross-section is : 31.7643 pb (indicative error: 0.518991 pb)
INFO: Original cross-section: 32.14466 +- 0.03739602 pb (cross-section from sum of weights: 32.144645)
INFO: Computed cross-section:
INFO: cHG : 31.7642987594 +- 0.51899110219 pb