Can I use the Chauvenet's criterion on set of observations where a normal distribution cannot be assumed? Thanks in advance!