I think it is primarily a matter of definition, where the choice of the definition you use is ruled first and foremost by pragmatic concerns.
As Ollett lays out in Chapter 1, there is a 'broad’definition that boils down to ‘all MIA that are not Sanskrit and not vernaculars’. And there is a narrow definition that takes into account the socisl, cultural and religious roles that were ascribed to the Prakrit languages by the Pre-Raj Indian society.
I would agree that dyachronical linguists should use the narrow definition. The text of the Ashokan pillars as well as the Pali canon belong to an era where Sanskrit as the codified and streamlined Vedic was still in its infancy (or maybe even non-existent yet). The Prakrit of Jain scriptures is undisputedly a product of a much later linguistic development than Pali or Ashokan. So, the use of Pali and Ashokan most likely pre-dates the tri-partite or bi-partite division of IA languahes by native grammarians. That means that historical linguistics would do well by steering away from simplistic and all too broad definition of Pali as Prakrit.
At the same time, for us as Buddhists , for the general public, etc. there is nothing wrong with using the broad definition. Most of the time they don’t need all the fine details if the linguistic history of India. Most don’t even realize that Sanskrit may be younger than Pali.
tl;dr: Pali is hardly a Prakrit but if you call it a Prakrit in a general context not requiring precision of expression it is fine.