Migration service using BulkMigrator should support cluster restart

Today the BulkMigrator processor (migration threads pool) is started when invoking a migration run operation.

A new worker node is not going to participate to the migration, restarting all worker nodes results in a migration that cannot complete because it is not possible to run the processor again, the migration stays in a running state

    {
      "entity-type": "migration",
      "id": "blob-keys-migration",
      "description": "Populate ecm:blobKeys property",
      "descriptionLabel": "migration.dbs.blob.keys",
      "status": {
        "state": null,
        "step": "empty-to-populated",
        "startTime": 1687342496120,
        "pingTime": 1687354769323,
        "progressMessage": "Migrating content",
        "progressNum": 0,
        "progressTotal": 12216,
        "running": true
      },
      "steps": []
    },

If we run the migration again we receive a 400 reply with:

IllegalArgumentException: Migration: blob-keys-migration already running step: empty-to-populated"

Also a thread is maintain on the front node that receive the migration request.

The only critical part where restart should be avoided is during the scroll of the bulk command used by the migration, the rest should be able to support a restart, in order to scale or to support a failure resolution when applying a hot fix.