Notes. Distinguishing the four different cases considered in Sects. 3.1 and 3.2 (ground-based vs. space-based simulated observations, and ideal vs. galfit deflector subtraction), we report in this table the mean bias and the scatter observed for the three main parameters of the lens model (R_Ein, μ, m_img), along with the rate of catastrophic errors defined as the systems for which | ΔR_Ein | /δR_Ein> 10, knowing that the errors are underestimated by the fast modeling. For the time aspect, we give the median execution time and the rate of failures corresponding to the cases where the MCMC did not converge within a threshold time relative to the median time (1 min for CFHT simulations regardless of the subtraction method, 2 min for HST simulations with an ideal subtraction, and 45 min for HST simulations with a galfit subtraction due to the necessary × 4 sampling while simulating and optimizing).