Make logical dump/restore stages more predictable between restarts

Goal

Keep the DLE retrieval component more predictable, especially between restarts

TODO / How to implement

• Skip running restore jobs(logicalDump and logicalRestore) if pgdata is valid (zfs snapshot already exists and clones can use it):
  • check if the data directory is not empty
  • check the value of data state at inside the data directory
  • refresh is running by timetable
• Cleanup the dumpLocation directory before taking a dump (logicalDump job)

Acceptance criteria