Allow multiple schedulers to run

[cheungpat]

Schedulers have to acquire a lock before one can schedule jobs. The
lock automatically expires in case a scheduler is terminated unexpectedly.
Schedulers that cannot acquire a lock will sleep for the duration of
polling interval before retrying.

refs #70

[selwin]

Thanks for the PR!

I'd like to keep register_birth to keep track of when the active scheduler instance taking over so could you please use a different key for lock acquisition purposes?

For backward compatibility reasons, rq-scheduler should still raise an exception if it fails to acquire a lock. The new behaviour should only kick in if the user specifies scheduler.run(retry_lock=n), with (-1) means retry for an infinite amount of time.

Better naming suggestion welcome :)

[selwin]

@cheungpat reminder to update this Pr when you have the time :)

[marcinn]

@selwin @cheungpat - any chances to finish this PR and merge into upstream? I'm building a failover in one of my project and RQ Scheduler is currently a SPOF. I can use external tool like Heartbeat, of course, but the possibility to make RQ Scheduler fault tolerant without external tools is attractive and simpler in terms of overall architecture.

Can I help you somehow with this PR?

[selwin]

@marcinn if you can take care of the merge conflicts, please do so and submit another PR :)