Implement new Web API endpoint /save/bulk/

This is a sub-task for implementing a MVP on the Bulk On-demand Archival feature.

To do so, a new Web API endpoint /save/bulk/ must be added.

It must be protected with authentication and special user permission.

User can submit a list of origins and their visit types (in CSV or JSON format) through a POST request:

• webapp is first performing basic checks on the URLs (is their format valid ?) and their visit types (are they supported by SWH ?) => return 400 if an issue is detected

• webapp then creates a scheduler task for the new save-bulk lister with parameters:

  • instance = user identifier
  • origins = a list of origins and their visit types (list of tuples or list of dicts)

  from swh.scheduler.utils import create_oneshot_task
        
  scheduler.create_tasks([
      create_one_shot_task("list-save-bulk", 
                           priority="high", 
                           instance=user_id, 
                           origins=origins)
  ])

  => return 200 if the task was successfully created

• user can track the statuses of the origins he submitted for archival through a GET request:

  • after the save-bulk lister was scheduled and executed, the list of origins validated by it and their loading statuses can be retrieved using the scheduler API

  from swh.core.api.classes import stream_results
  lister = scheduler.get_lister(name="save-bulk", instance_name=user_id)
  listed_origins = list(
      stream_results(scheduler.get_listed_origins, lister_id=lister.id)
  )
  origin_visit_stats = scheduler.origin_visit_stats_get(
      ids=[(o.url, o.visit_type) for o in listed_origins]
  )

  • info about origins rejected by the lister (404, invalid visit type, ...) can be found in the lister state (for the MVP case, ideally such info should be stored in a database table not as a JSON blob in lister table)

  lister = scheduler.get_lister(name="save-bulk", instance_name=user_id)
  rejected_origins = lister.current_state["rejected_origins"]

  • Web API should return a JSON response with some loading statistics about submitted origins but also detailed info about processed and rejected origins (maybe in an on-demand way using a query parameter ?)

Related to swh/meta#5091.