A lot of things are being discussed here regarding piranhas and pacus. Some are supported by the science, others are not. A reading of this relatively recent paper by Kolmann et al 2020 (DOI:10.1093/sysbio/syaa065) may help dispel or clarify the context for some of those questions about members of the family Serrasalmidae -
1) Likelihood that piranhas and true pacus will interbreed. Of course not discussed per se, but by examining the reconstructed tree it can be interpreted to be unlikely. The piranha genera (Subfamily Serrasalminae) are relatively far removed phylogenetically from the pacu genera (Subfamily Colossomatidae). There are likely genetic barriers and others so that even artificially it would be unlikely.
If anything, Metynnis silver dollars are the most closely related SD's to the piranhas, being in the same subfamily group (Serrasalminae).
2) Shape of teeth as indicative of phylogeny and of dietary habits within the whole family Serrasalmidae -
The overall conclusion is that there is enough parallelism among diverse genera and species onto common diets, and enough convergent evolution, such that diet is not a good indication of tooth morphology within the family. There are both cutting teeth, and well as molariform examples within the piranhas, and also there are clearly sharp teeth among some non-piranha, or non-carnivorous species. And yes, there probably is some degree of dietary shifts during development of some species.
That does not argue against the famous piranha-monster pictures, which will always have a very sharp toothed species shown. But it does not always allow to generalize that a single morphology is common to all species of a given group.
I believe the paper is available under Free Access at the link above (copy and paste). I post the abstract below.
