Assessing the impact of discretisation assumptions in a length-structured population growth model