Tabula rasa for erfcx integral?

[arashgmn]

Numerical integration of the erfcx is hard. But since all mean-field needs is a definite integral, isn't it better to numerically solve this integral once, with very high precision, from $-\infty$ to $x$ and then evaluate the finite integral using interpolation (for the limits) and the fundamental theorem of calculus? To me it seems to speed up the code and also lower the numerical difficulty of this line in the delta synapse

nnmt/nnmt/lif/delta.py

Line 155 in a5cc295

def _firing_rates_for_given_input(mu, sigma, V_0_rel, V_th_rel, tau_m, tau_r):

and its equivalent line in the exponential synapse.