redo LAI exp of nleaf + correct bug to consider decrease LAI following senescence in that case (SIM) + modify optimization to compute delta LAI following these changes

improve consistency between comp_simul*py and plot_multi_simul*py + add filter within plot_multi_simul*py

correct bug related to division by SD and precision (leads to some issue in Sobol) following carbo LAI based computation (instead of nleaf) + improve test for checking

progress in plot of Tartas GPCROP simulations