On the suitability of thermogravimetric balances for the study of biomass pyrolysis