What is in this PR:

1. Improve the file management in Operandi:

• Introduce an extra download worker with a new download queue. The job status worker pushes download requests to the download queue if the status of a job becomes SUCCESS. Unlike the status queue, the download queue is persistent and survives Operandi service restarts. Hence, downloading HPC results (workflow jobs/workspace results) would not be interrupted.
• Refactor all worker classes in the Operandi Broker to remove duplicate code and simplify the job status worker. The downloaded HPC results of succeeded jobs are now properly deleted from the HPC. The failed jobs for the current time being are still preserved only in the HPC and not downloaded (still to be decided how to handle files of failed jobs properly)
• Zips to be transferred/downloaded to the HPC environment are now properly deleted from the Operandi Server local storage after being transferred to/from the HPC. Previously, some zips were not deleted under certain circumstances (e.g. exceptions due to wrong zip format or failing to unzip).
• Set the logging level of some messages to DEBUG of the OtoN converter to prevent extensive logging flooding
• Reduce log levels of some 3rd party libraries (pika, paramiko, ocrd-core) to WARNING to prevent extensive logging flooding
• Other small improvements and fixes

2. Add a flag for the admin batch endpoints to let the admins hide/show deleted resources when requesting resources. Regular users can see only non-deleted resources. Usually when a resource is deleted, the deletion happens locally but the database still keeps all the metadata of that record. The deleted resources in the DB are identified with the deleted field of each resource.