Class: JobQueueRedis  - X-Ref

Class to handle job queues stored in Redis

This is faster, less resource intensive, queue that JobQueueDB.
All data for a queue using this class is placed into one redis server.

There are eight main redis keys used to track jobs:
- l-unclaimed  : A list of job IDs used for ready unclaimed jobs
- z-claimed    : A sorted set of (job ID, UNIX timestamp as score) used for job retries
- z-abandoned  : A sorted set of (job ID, UNIX timestamp as score) used for broken jobs
- z-delayed    : A sorted set of (job ID, UNIX timestamp as score) used for delayed jobs
- h-idBySha1   : A hash of (SHA1 => job ID) for unclaimed jobs used for de-duplication
- h-sha1ById   : A hash of (job ID => SHA1) for unclaimed jobs used for de-duplication
- h-attempts   : A hash of (job ID => attempt count) used for job claiming/retries
- h-data       : A hash of (job ID => serialized blobs) for job storage
A job ID can be in only one of z-delayed, l-unclaimed, z-claimed, and z-abandoned.
If an ID appears in any of those lists, it should have a h-data entry for its ID.
If a job has a SHA1 de-duplication value and its ID is in l-unclaimed or z-delayed, then
there should be no other such jobs with that SHA1. Every h-idBySha1 entry has an h-sha1ById
entry and every h-sha1ById must refer to an ID that is l-unclaimed. If a job has its
ID in z-claimed or z-abandoned, then it must also have an h-attempts entry for its ID.

Additionally, "rootjob:* keys track "root jobs" used for additional de-duplication.
Aside from root job keys, all keys have no expiry, and are only removed when jobs are run.
All the keys are prefixed with the relevant wiki ID information.

This class requires Redis 2.6 as it makes use Lua scripts for fast atomic operations.
Additionally, it should be noted that redis has different persistence modes, such
as rdb snapshots, journaling, and no persistent. Appropriate configuration should be
made on the servers based on what queues are using it and what tolerance they have.