Fitting insurance claims to skewed distributions: Are the skew-normal and skew-student good models?