fix DLE restore configuration
With default settings, DLE's sync instances doesn't handle failover properly (Postgres versions < 12): add recovery_target_timeline = 'latest'
GitLab.com database experienced a failover event at ~2021-01-16 23:40, and the state of the Database Lab Engine's sync instance became frozen, not being able to progress further.
The root cause is the insufficient configuration of the sync instance -- lack of this line:
recovery_target_timeline = 'latest'
This setting became default in PostgreSQL 12, but GitLab.com is on Postgres 11.
Once this line was added to recovery.conf, and Postgres in the sync container restarted, switching to a new timeline was complete, and replaying WALs continued.
We need to adjust the default configuration of DLE to handle failovers properly in Postgres 11 and older (we can have that line explicitly defined in configs for Postgres 12+ as well).
Also, it is necessary to correct the path of the recovery file and skip an error due to absent
postgresql.conf while init DLE configs