We should define the Order of Operations for Operator Reconciliation

In https://docs.couchbase.com/operator/current/concept-operator.html we should outline the order of operations that Operator takes when it performs reconcilliation.

This should be granular enough that a customer who makes multiple changes to their cluster, should be able to predict the order the changes will be applied.

As an example:

Customer changes vCPU limits/request from 5 to 15, and scales down the cluster from 16 to 6.

Right now, there is no documentation that would indicate to the user that we would first increase the vCPU of the 16 pod cluster, and then scale down to 6 nodes. This caused a customer's cluster to fail an upgrade (as they did not have enough resources available to do the vCPU scale first, then the scale down)