I like to think of the distinction in a de jure vs de facto way.

If all the evidence we have of a species is them consuming autotrophs, then we do mental gymnastics and induce that they are 100% herbivores.

Of course this also leaves room for an herbivore’s potential to consume heterotrophs, in which case our knowledge would have to update and reflect reality. Maybe 99% herbivore, 1% carnivore.

And at that point, we may still do mental gymnastics and call species herbivores because that is their normal behavior, where their abnormal behavior is shown due to abiotic or biotic factors, perhaps from loss of habitat or removal/introduction of species in the food web, etc.