Geo queues not drained when turned off

Customer noticed lots of Redis memory was being used. The Redis bigkeys analysis showed:

Biggest   list found 'geo:gitlab:updated_wikis' has 112704799 items
Biggest   zset found 'resque:gitlab:cron_job:geo_bulk_notify_worker:enqueued' has 5473 members

They had Geo running for 6 months but turned it off. Looks like a number of queues were never cleared out. Do we need some way to drain these queues?

ZD: https://gitlab.zendesk.com/agent/tickets/82061