Fix a couple run groups not exiting

[RebeccaMahany]

Relates to #1205

A few of of our 12 rungroup actors would not return from Execute on Interrupt, locking shutdown:

• TraceExporter was waiting for a cancel that wouldn't happen until after the rungroup exited -- fixed to use the usual interrupt channel instead
• LogShipper (via sendbuffer) would not exit on interrupt because we were breaking out of a select and not the for loop containing it
• Updater (for both launcher and osquery) was, similar to the TraceExporter, waiting for a cancel that wouldn't happen until after the rungroup exited -- to minimize the changeset in this area of the codebase, I just made a new context and ensured we call the cancel on interrupt

I also made some improvements to how quickly the new TUF autoupdater will respond to shutdown requests. If it's in the middle of an update, it can take a few minutes failing and retrying when querying the current running version of osquery (since the osquery extension has already shut down). Now we check availability before checking current running version, since that's a faster short-circuit, and we perform the osquery retries more quickly.