On the viability of the single-wave model for the beam plasma instability