Auto-scaling follow-up

The 2.2 documentation for auto-scaling is extensive, but some tasks and behavior are implied and not explicitly divided out into sections, which may or may not affect discoverability. Some functionality also wasn't explicitly documented because it wasn't tested yet.

This ticket tracks any follow-on subjects that need to be documented for 2.2 auto-scaling.

The following is a running list of subjects that should be considered:

1. How does auto-scaling behave during Couchbase upgrade (e.g. does HPA go into maintenance mode)
2. How does auto-scaling behave during Operator upgrade (e.g. does auto-scaling temporarily pause)
3. How does auto-scaling behave during different types of metrics disruption (e.g. Prometheus or Metrics Server crashes, network timeouts, etc.)
4. How to adequately configure auto-scaling for a server class with more than one service (e.g. Index/Query)
5. How to safely modify your auto-scaling configuration (e.g. edit HPA)
6. How to safely “turn off” auto-scaling for a cluster and why you might do it (e.g. I was using auto-scaling before, but I don’t want to use it anymore)