Safer bootstrapping

[paweljw]

• wait for up to 30s during bootstrap for Beyla socket to become available (assume slow start)
• if it doesn't, start the local stack as-is, write a marker file denoting unsuccessful Beyla bootstrap
• a change in remote files in telemetry (pending) will look for this marker and try to re-provision Beyla container if the socket becomes available
• more guardrails around bootstrap:
  • if bootstrap marker is written,
  • but updater isn't running,
  • and bootstrap.sh is ran again,
  • assume bad bootstrap (maybe a bad file was pushed from telemetry); delete markers and try again

Additionally bumps Beyla to 2.7.7 - it's not significant for us, but since I'm cutting a release anyway, might as well.

[curusarn]

Thanks!

Intuitively I'd hope Beyla could bootstrap/restart async. But this looks ok.

[paweljw]

@curusarn I'm still thinking about that route, but it feels more brittle - more authority spread = more timing issues possible. We'll see :)