Microservices are 100% in the "Optimize for flexibility and you'll end up with something that's way too complex" territory.
(I am watching this right now at my workplace: team next to us has updated the microservice in repo A, and accidentally made incompatible change, so that integration tests in repo B which make calls to that microservice are now falling. Is it a solvable problem? Certainly. But back before we had microservices, this problem just could not occur, as client/server versions were always identical)
They are definitely not autonomous then. Seems like a distributed monolith.
Micro-services are supposed to be simple and independent, but they get crammed with more and more features and complexity and suddenly they just can't be maintained without breaking other parts of the architecture.
Yeah, I've only seen microservices used in a couple of places, but in both cases I felt they were introduced far too early. The first time I saw them was for a 2 page note-taking web app, and it had something like 65 microservices.
(I am watching this right now at my workplace: team next to us has updated the microservice in repo A, and accidentally made incompatible change, so that integration tests in repo B which make calls to that microservice are now falling. Is it a solvable problem? Certainly. But back before we had microservices, this problem just could not occur, as client/server versions were always identical)