Reconsider DeadlineExceeded error #15313

Currently, we raise DeadlineExceeded if the overall operation times out, with the retries exposed as a RetryExceptionGroup attached.

Is this the right long-term decision for GA? This is a bit custom, and not how other GCP libraries handle this

Maybe a better solution would be to make a separate OperationTimeout exception, completely separate from the grpc deadline error. We can make it a subclass of RetryError in api_core, and make all libraries align on this

see #15066