Add info about things in memory being lost after suspend/stop

[dacharyc]

When Tugboat's Preview containers are stopped, which happens when a user manually stops a container and when Previews are suspended - things that are contained in memory at the time of the stop are lost.

This doesn't appear to be stated explicitly in the docs anywhere, so we should add a note around this. Ben suggests this might make sense as part of the [set of] preview lifecycle diagram[s]. Consider where is the best place to add this information, and possibly make it a callout somewhere.

[q0rban]

One thing to stress here is that (I think) this is a Docker thing, not a Tugboat thing. It's not really anything we have control over.

[bchavet]

I don't think it's important to make a Docker/Tugboat distinction here. The purpose of suspending a preview is to recover unused resources in our shared environment. This lets us host far more previews than could be running simultaneously.

We chose to do this by stopping the service containers. We have no way to know how long a preview will remain suspended. Stopping the containers is the most reliable way to ensure a consistent state, to survive host server reboots, service migrations to other hosts, infrastructure maintenance/outages, etc.

In our implementation, a Stopped preview and a Suspended preview are identical, except that a Suspended preview will automatically start when it receives a request. This also makes things more straightforward to users maintaining their build scripts. There is only one Start case to consider, whatever that might mean for their project.