Add Redis connection health checks and restarts

[mjeffrey18]

Redis has complexities that need to be considered;

1. Redis Pub/Sub works well until you lose the connection...
2. Redis connections can go stale without activity.
3. Redis DB's have a buffer related to the message sizes called Output Buffer Limits. Exceeding this buffer will not disconnect the connection. It just yields it dead. You cannot know about this except by monitoring logs/metrics.

This PR helps to mitigate issues 1 and 2

We've been running these changes in production (with these changes #54 and a different Redis shard) with excellent results;

1 month of APM metrics

NOTE: No automated restarts at all in the last month

Workloads

NOTE: Those errors are all just from Lucky No route (spam requests)

2 Servers will low resource usage

[mjeffrey18]

@jwoertink is ready for initial impressions review. I still have to add specs. I will sort that out tomorrow.
I've changed some of the logging to match your PR #48

[jwoertink]

Thanks @mjeffrey18! @russ and @franciscoGPS if either of you have time, I'd love more eyes on this.

[russ]

This looks awesome. Can’t wait to try it out.

[mjeffrey18]

This PR plus - jgaskins/redis#35 is what we are running in production

[mjeffrey18]

@jwoertink Added specs for this also. We can merge if you guys are ready.

[fernandes]

adding this section on README is amazing, thanks for such effort 😎 👍

[jwoertink]

Sweet! Thanks for doing this. Sorry for the delay, I was at Euruko last week. Just got back and catching up.