Docs should highlight write availability issue with durable writes and 1 replica buckets

When a node fails in a 1 replica bucket, durable writes against vbuckets on that node can't run to completion. This happens because durable writes need to be acked by a majority of the available copies of the data. A 1 replica bucket has 2 copies of every vbucket and a majority of 2 is 2 – so the replica needs to be available for durable writes to succeed.

This is a known issue and is captured in this table in the documentation: https://docs.couchbase.com/server/current/learn/data/durability.html#majority.

However, it's a trap that quite a number of users & customers have fallen into in the past, especially since 1 replica buckets are the most common way that Couchbase Server is deployed.

We should add a special "warning" or "important note" that if a node fails and a bucket is configured with one replica, durable writes will be partially unavailable (for any vbucket that lives on the failed node.)