After performing the checks and starting to acquire the database critical section, the create database coordinator needs to always make progress.

We will still want to keep the limit on the number of retries on ShardNotFound error. The coordinator should then abort and clean up (release the critical section) after exhausting the retries.