Generic Executor health checker that uses signals independent from deployment environment

[eabatalov]

Current Executor health check policy and reasoning:

• A Function Executor health check failure is a strong signal that something is wrong either with:
  • The Function Code (a criticial software bug).
  • The Executor machine/container/VM (a software bug or malfunctioning local hardware).
• Critical Function Code bugs tend to get fixed eventually by users. What doesn't get fixed eventually is rare but recurring local Executor issues like hardware errors and software bugs in middleware like drivers.
• Such issues tend to get mitigated by automatically recreating the Executor machine/VM/container.
• So we fail whole Executor health check if a Function Executor health check ever failed to hint the users that we probably need to recreate the Executor machine/VM/container (unless there's a bug in Function code that user can investigate themself).

Also implemented tests for both startup and health probes.