Further clarify where to point xpack.monitoring.elasticsearch.url to

*Original comment by @ppf2:*

We previously added the clarification on not pointing xpack.monitoring.elasticsearch.url to a separate monitoring cluster even if there is one, but point it to the main cluster.

It will be helpful if we can further improve the documentation on xpack.monitoring.elasticsearch.url with what we will advise users to set xpack.monitoring.elasticsearch.url to for the main cluster. For example, something like the following:

>For xpack.monitoring.elasticsearch.url, you can specify an array of hosts and it will attempt to round robin the connection until it finds one that is responding.
>- If you have coordinating-only nodes in the cluster, point xpack.monitoring.elasticsearch.url to the array of client nodes you have in the main cluster
>- If you do not have coordinating-only nodes in the cluster, point xpack.monitoring.elasticsearch.url to the array of data nodes you have in the main cluster
>- Do not include dedicated master nodes in the pack.monitoring.elasticsearch.url array.

Also, on the https://www.elastic.co/guide/en/x-pack/5.2/monitoring-settings.html page, it says "instance(s)" which is a way of saying that the setting accepts a list of hosts. It will be nice to provide a quick LINK REDACTED array syntax for completeness syntax for it.