Improve scaling documentation for Streaming

[ThisIsMissEm]

This reorganizes the documentation a little bit, add additional information about:

• the templated systemd files & how to start multiple instances
• information about the Prometheus metrics endpoint, and some key metrics to observe.
• Some reference numbers for concurrent streaming connections (this is based on my experience)
• Moved the warning about STREAMING_CLUSTER_NUM to the bottom of the section (since it's now two versions out of date at least).

[vercel]

@ThisIsMissEm is attempting to deploy a commit to the Mastodon Team on Vercel.

A member of the Team first needs to authorize it.