`GenerateLISATDI_SMBH` does not honour the frequency cut

The source code is to find in https://gitlab.in2p3.fr/marsat/lisabeta/-/blob/master/lisabeta/lisa/lisa.py?ref_type=heads#:~:text=%23%20If%20the%20user,EOBNRv2HMROM.%27).

    # If the user specified gridfreq, force to use it -- overriding other bounds
    # in frequency (e.g. minf, maxf) or time (e.g. DeltatL_cut)
    input_gridfreq = not gridfreq is None
    if not input_gridfreq:
        # Determine frequency bounds based on frequency and time limits
        fLow, fHigh = FrequencyBoundsLISATDI_SMBH(params, t0=t0, timetomerger_max=timetomerger_max, minf=minf, maxf=maxf, fstart22=fstart22, fend22=fend22, tmin=tmin, tmax=tmax, Mfmax_model=Mfmax_model, DeltatL_cut=DeltatL_cut, DeltatSSB_cut=DeltatSSB_cut, scale_freq_hm=scale_freq_hm, modes=modes, approximant=approximant)

        # Combined frequency grid for the waveform
        # If HM and not using m/2 scaling, do log-affine scaling
        # with frequency bounds for individual modes
        if isinstance(fLow, dict) and isinstance(fHigh, dict):
            gridfreqClass = pytools.FrequencyGrid(fLow[(2,2)], fHigh[(2,2)], M, q, acc=acc, DeltalnMf_max=DeltalnMf_max)
            gridfreq22 = gridfreqClass.get_freq()
            gridfreq = {}
            for lm in modes:
                gridfreq[lm] = pytools.log_affine_scaling(gridfreq22, fLow[lm], fHigh[lm])
        else:
            # For PhenomHM will be rescaled by m/2 for different modes hlm
            gridfreqClass = pytools.FrequencyGrid(fLow, fHigh, M, q, acc=acc, DeltalnMf_max=DeltalnMf_max)
            gridfreq = gridfreqClass.get_freq()
    else:
        # NOTE: when loading from json, gridfreq will be a list
        # gridfreq can also be a dictionary of numpy arrays, one for each mode
        if isinstance(gridfreq, list):
            gridfreq = np.array(gridfreq)
        if (approximant=='EOBNRv2HMROM'):
            raise ValueError('Input gridfreq not supported for EOBNRv2HMROM.')

FrequencyBoundsLISATDI_SMBH makes sure the frequency cut-off (minf, maxf) is applied to all the harmonics, but this information is ignored in the rescaling process. This is unwanted, right?