Some Considerations on the Choice of a Probabilistic Model for Reliability Assessment of Power System Components