Slit dispersion bug over large spectral range

[CorentinGrimaldi]

🐛 Describe the bug

When applying a slit function with a slit dispersion function to a spectrum calculated over a large spectral range, there is a bug in the quantities modified by the (changing) slit.
This is specially visible for the transmittance plot.

This is probably due to an error with the gestion of the boundaries during the convolution of the slit function with the spectral quantity.

🖥 Steps To Reproduce

Calculate a spectrum, for example H2O between 1000 and 1100 nm, apply a slit function whith a slit dispersion function and plot any quantity modified by the slit function. The bug is particularly visible for the transmittance.

from radis import calc_spectrum
import numpy as np

def slit_dispersion(w):
    phi = -6.33
    f = 750
    gr = 600
    m = 1
    phi *= -2*np.pi/360
    d = 1e-3/gr
    disp = w/(2*f)*(-np.tan(phi)+np.sqrt((2*d/m/(w*1e-9)*np.cos(phi))**2-1))
    return disp  # nm/mm


spec_h2o = calc_spectrum(wavelength_min=1000,
                         wavelength_max=1100,
                         Tgas=300,
                         molecule='H2O',
                         isotope='1',
                         )

spec_h2o.apply_slit(1, slit_dispersion=slit_dispersion)
spec_h2o.plot('transmittance')

💡 Possible solutions

Change the mode in the apply_slit() function.