Be clear about the auto-rename that can occur when node added to single node cluster

All clusters begin life as a single node. This node is often provisioned using the node name 127.0.0.1 as:

1. this is the default and
2. when a second node is added, ns_server automatically renames the first node to a non-loopback interface in a way that is often (but not always) convenient for users.

This automatic rename only happens when the first node is configured on the loopback address (127.0.0.1 or ::1) and of course it's necessary to do the automatic rename when a second node is added as it's impossible for the second node to contact the first node if the first node advertises its name as a loopback address.

The auto rename is convenient because it means users don't need to figure out which interface to use and carefully type it in and people like conveniences. The problem is that sometimes the server doesn't pick the address that the user wants.

We should document the mechanism by which the server auto renames so that users understand what's happening and know how to avoid situations in which they get unexpected results.

The mechanism is:

• The cluster node (the one configured with the loopback interface) opens a connection to the second node via the name provided and determines the interface used for the source port on the connection
• The cluster node renames itself to this interface that it has discovered

Cloud / split DNS horizon environments are particularly prone to users not getting behavior they expect as frequently nodes will be named with public DNS names and yet communicate with each other via non-routable IP addresses.