Implement pure NodeJS jobs queue backed by Bull

[joseconstela]

The current implementation of the jobs queue (the engine in charge of running and orchestrating workflows executions) is built in top of msavin:sjobs-ui-blaze, which at the same time is built in top of MeteorJS.

The new Jobs queue service should be created from the scratch, not using either msavin:sjobs-ui-blaze or MeteorJS. This will allow to:

• Run the queue either as a separate process, or within the same process as Tideflow's
• Run the queue in a separate machine or server
• Let the jobs and Tideflow itself to use separate database connections
• Do not depend in possible obsolete or unmaintained MeteorJs packages.
• Possiblity to fork queues processes for blazzing fast performance.
• To have a separate dashboard to see metrics on the queue.

To be discussed:

• How to keep modularity? as the services will run on the queue, but they also have visual elements that are rendered on the main application.

---

Subtasks

• Skeleton with MongodDB connection
• Rewrite job workflow-start
• Rewrite job workflow-step
• Rewrite Eliminate job workflow-execution-notify-email
• Rewrite job workflow-execution-finished
• Rewrite job _executionLogsRun
• Rewrite job _executionLogsSendEmail
• Rewrite job s-cron-runOne
• Rewrite job s-rss-runOne
• Rewrite job s-rss-schedule
• Move the service's server/service to the new queue