Ability to throttle jobs

[Blacksmoke16]

Resolves #1

• Ability for arbitrary limits/periods.
• Config hash could be used for other things, like retry amounts etc
• Default is no limit
• ~Maintains state between starts, would have to delete the old config hash to have it set new values~
• Renames the WAITING queue to actually be waiting, is more clear.
• Refactors the queue methods into a nice macro :)

~However the more i think about it the one thing this wouldn't solve well is if the tasks get added slowly. For example you have a job that is set to 10 executions per 60 seconds. But you add one every 10min. This implementation would still would wait 60 seconds before executing the 11th task.~

This has been solved, it now only counts executions within the given period seconds.

How this works:

1. I moved the @@mapping class var to a class_getter var, as i needed a way to get a list of all defined jobs
2. When starting mosquito it will store a config hash into redis
3. When a task is executed it will increment the executed field in the config object
4. If Time.utc_now.epoch > (Time.epoch(config["last_executed"].to_i64) + config["period"].to_i.seconds).epoch resets executed back to 0
5. When executed == limit, it will set the executed field back to 0 and set the next_batch field to (Time.utc_now + period.seconds).epoch
6. When going to dequeue a task to run, it checks q_config["next_batch"].to_i64 > Time.utc_now.epoch and returns if true

WDYT @robacarp

[Blacksmoke16]

~Noticing an issue where its doing 1 more execution than it should sometimes, i'll look into it.~

EDIT: Fixed in b4930ce

[greenbigfrog]

Any reason why this hasn't been merged yet?

[Blacksmoke16]

IIRC I need to write tests for it, but was dependent on another shard being updated, that has since been forked and update.

I been using it for past few months and it's been fine as far as i can tell. However, not sure on the best way to test this, and hasn't been a high priority of mine lately :/

[Blacksmoke16]

@robacarp This should be good for review again when you get some time.