This PR introduces following changes:
1) Fix race condition where `swaps_configure` can be called when another call to the same endpoint is already executing resulting in undesired behaviour.
2) Do not try to `savecore` for partitions which are already being used in swap.
3) Configure dump device again if current dump device is removed from swap.
4) Fix race condition where sometimes creating a pool would not result in swap being configured as we rely on zfs event to configure swap which is raised as soon as the pool is created. However, during configuration we use database to find out current pools and retrieve valid disks - but if the pool hasn't been saved yet in the database, we don't find those disks resulting in swap not being configured.