Why can't you have both independently deployed microservices and a shared code base?
If the deployment lifecycle is different for each microservices and each deployment is self-contained, then they can be deployed with different versions of the code - even if they use the same source tree and share code.
Obviously the shared code needs to be properly maintained and evolved, but it seems to me a lot of the software engineering problems occur when people move away from source code dependencies - with great tooling - versioning, diffs, debuggers - to other types of dependencies ( shared libs etc ) where the tools are non-existent or very simple.
Now granted if you needed to fix a critical bug in the shared code - that would require a redeploy of everything, but that happens much less frequently than the need to deploy a single service with immunity as long as your keep your microservice contract. It also means the discipline of making sure every services is deploy-able at anytime is kept to.
And if you didn't share code - you probably wouldn't be fixing a single bug once, you'd have much more code, with many more bugs.