The finished jobs list (rq:finished:) never gets cleaned up

[cosminstefanxp]

First of all, great job with the project. It's awesome.

Now, to get to my problem. I'm using RQ in a high throughput system, with hundreds of thousands of jobs/day. Even though we have a result_ttl set up and the keys for the jobs themselves get cleared out, the rq:finished:low ZSET which contains ids of finished jobs does not seem to be cleared. As a result, since we put the system in production (4 days ago), it has grown to ~3M entries and keeps growing. This is starting to use a lot of memory and has a visible effect on the performance.

@nvie

[selwin]

You could cleanup this by calling finished_job_registry.cleanup(). We should also create a rq cleanup management script that can be called periodically to run cleanup jobs of all registries.

[selwin]

@cosminstefanxp please see #534 , this PR should address this issue.