Add timeout to osqueryinstance healthchecks

[zackattack01]

Some logs today showed the following on a windows device running launcher v1.12.3:

• all osqueryinstance errgroup routines starting correctly
• eventually, osquery stops logging and localserver started logging the following errors:
  • "msg":"updating id fields","component":"localserver","err":"id query failed: could not query the extension manager client: i/o timeout"
• this state continued for days, without the healthcheck routine triggering an osquery restart (or logging any sort of failure)
• when a shutdown was finally issued, the healthcheck goroutine never returned

I think the most likely explanation here is that it got caught up somehow, and that we should enforce some sort of cancellation on our Healthy() calls. This way we will at least log the failures and restart to eventually recover from this state