I know what you're saying, and I think this is why lots of people are abandoning personas in favor of things like JTBD. But I think personas (or something similar) can be useful for those other parts of the process, especially if they're based in research and aren't "imaginary".
For example, thinking about how a particular user flow would work for a specific relevant persona/JTBD is useful for design decision making. It helps you see things from a user archetype's eye, rather than through your own designer eyes or through the eyes of "the user" which isn't specific or helpful.
This is all assuming, of course, that you're building personas correctly, which in my experience is rarely done.
Personas might be useful as a brainstorming session to identify pain points, product interaction context, and problems to be solved. However, once those key goals have been identified, why not simply abandon the persona concept for the rest of the process?
Focus on solving the problems you've identified, not appealing to 3-5 extremely specific imaginary personas.