This was exactly my thought! Despite all the hoohah around the decisions made and whether or not they did things correctly, this idea of "we'll create a bunch of separate services and then use a common shared library for all/most of them" was the start of the end from where I'm sitting... doing this is exactly where the trouble starts with future code changes as the shared library almost becomes a god-like object amongst the services using it: change something in the shared lib and all services using it need to be re-tested. Unless proper versioning takes place, but that, from my experience, seems to rarely be the case. Or need proper service ownership and chain or notification to inform service owners when particular versions are being deprecated or retired... which seems to rarely be the case as well.

Even so, imagine the chaos if frequently engineers/devs need to add code to one lib(the shared one), wait for PR approval, then use that new version in a different lib to implement the actual change that was needed? Thats seems to be introducing a direct delay into getting anything productively done...