Job store support

[imiric]

This might be out of scope for schedule, but I think it would be very handy.

In a similar vein to what APScheduler has, we should support at least some kind of persistent job store.

Let's leave RAM as the default, but add an SQLite store as an option.

Reasonings:

• SQLite API is built into Python 2/3 and is rock-solid, so no additional dependencies.
• Keeps persistent state across restarts, as mentioned in Schedule jobs to run on specific weekdays #5.
• Ability to manage jobs from outside of the main schedule process, which avoids interrupting the process for that.
• Related to the above, ability to see when the next run is scheduled for (perhaps there could be a schedule CLI tool?).
• Centralizes the jobs so in theory we could daemonize the schedule process and have it manage all jobs on a single machine.

Thoughts? I might implement this myself soon, just wanted to see if there's any interest for it.

One particular hurdle I'll have to look into is how APScheduler is dealing with persisting jobs. I'm not familiar with pickling/storing functions.

[dbader]

I agree that this might be out of scope for schedule and it might also really hard to do job persistency correctly in all cases. Things like cron have been around for a long time and took a while to mature.

Still, I don't want to discourage you – this sounds like an awesome project that I'd definitely mention in the FAQ / future docs. So I say go for it and keep us posted about your progress :-)