Weights in the acceptance parametrisation

According to the code, there's the possibility to specify a weight variable (weight-var-name) in the config file used to get the acceptance shape of a given sample. When using this option, however, I observe something strange in the results (see attached plots).

This is the plot I obtain by doing:

python model_efficiency.py ~/LHCb/Analysis/Repository/ewp-kstee-fullangular/Software/Acceptance/B02Kst0ee/Conf-GetAcceptance-B02Kst0ee-Rec-1To7All-L0E-4D-Corr-5nq2-4nctl-6nctk-5nphi-0ctleven-0phieven-wTot.yaml

where the config file is the following:

name: B02Kst0ee-Rec-1To7All-L0E-4D-Corr-5nq2-4nctl-6nctk-5nphi-0ctleven-0phieven-wTot 
model: legendre
data: 
    source: /disk/data12/lhcb/flionett/LHCb/Analysis/AnalysisResults/RareMode/SamplesForMassFits/SamplesWithTrigEffCorr/SamplesWithTotalWeightForAcceptance/B02Kst0ee-RunI-MC-central-L0E-0.85-HOP0-BkgCat1-Ctl0-51500-Mass4500To5800-L0HLTrw-wTot.h5
    tree: B02Kst0ee
    weight-var-name: wTot
variables:
  - q2_PVandBDTF
  - ctl
  - ctk
  - phi
parameters:
    legendre_orders:
        q2_PVandBDTF: 5
        ctl: 4
        ctk: 6
        phi: 5
    ranges:
        phi:
            - -pi
            - pi
        q2_PVandBDTF:
            - 1.1 
            - 7.
plot: y
plot-labels:
    q2_PVandBDTF: '$q^2$ (GeV$^2/c^4$)'
    ctl: '$cos\theta_\ell$'
    ctk: '$cos\theta_K$'
    phi: '$\phi$ (rad)'

At first I thought it might have been a problem with the weights, so I plotted the q2 distribution with and without the weight wTot, which is the following:

Based on this, I would say that there's something strange with the acceptance part, do you @apuignav and @jeschle see anything unusual? Was this weight option tested somehow?