Document DR steps for Geo in Kubernetes

Problem to solve

When using Geo in k8s with our Helm charts, specific steps might be different from our general DR docs. We should investigate and document failover steps using our Helm charts.

Further details

Specific points that may come up:

• Whether we need to run everything as in the general DR docs, and where we need to run them (task-runner pod)
• Configuration changes, differing from the general DR docs (values.yaml vs gitlab.rb)
• Document handling the promotion steps on external (from k8s) Omnibus databases

Other links/references

Came up in a Slack thread (internal only) discussion.

/cc @fzimmer @hchouraria