RQ Worker timeout doesn't yield an error

[torgeirl]

If a batchframework_admin_compress_assignment operation fails due to a timeout caused by the task to build a Tar archive file exceeding the timeout set by RQ_QUEUES['default']['DEFAULT_TIMEOUT'], the failure is only reflected by the status in the Django admin UI.

Status: finished
Result: failed
Output: 
{"actiongroup_name": "batchframework_admin_compress_assignment",
  "results": [{
    "action_name": "devilry.devilry_admin.tasks.AssignmentCompressAction",
    "failed": true, "errordata": {
      "data": "An HTTP Client raised an unhandled exception: Task exceeded maximum timeout value (500 seconds)"
      ...

The timeout does not yield an error (so it could be detected by Sentry), and the RQ worker itself just logs «Job OK»:

[INFO] rq.worker: default: ievv_opensource.ievv_batchframework.rq_tasks.default(
    actiongroup_name='batchframework_admin_compress_assignment',
    batchoperation_id=<batch-operation-ID>,
    operationtype='batchframework_admin_compress_assignment') (<job-hash>)
[INFO] rq.worker: default: Job OK (<job-hash>)

[torgeirl]

Partly addressed in b868acb (doesn't cover timeout, but issues like #1322).