Update UnifiedPush SOP to describe how to handle memory alerts.

The current UPS SOP refers to scaling the pod horizontally if a memory alert is issued. This is incorrect and the SOP should instead describe how to redeploy the pod without causing downtime.

Link to SOP in question : https://github.com/aerogear/unifiedpush-operator/blob/330f9f7ffd62a02fd4d0d78c16cfed7c328c1e5d/SOP/SOP-push.adoc#unifiedpushjavaheapthresholdexceeded