Simulations and multidimensional poverty: reflections on a COVID-19 evaluation study