Some guidelines for a consistent use of the Weibull statistics with ceramic fibres