On the validity of the cluster model to describe the evolution of Cu precipitates in Fe–Cu alloys