On the impact of the biasing history on the characterization of random telegraph noise