Provide way to auto recreate promscale pod in some situation

[zfy0701]

What is missing? Currently, if promscale can't connect to timescaledb, it tries to reconnect but it may never succeed unless manually deleted the promscale pod

the above situation happens when tsdb get restart (but not every restart triggers this)

we also notice if tsdb restart with no promscale extension and if promscale is already started, promscale will complain no promscale extension is installed, but it will never restart, which will fix the problem.

Why do we need it? I assume we could provide a liveness check configuration so we can curl and check on a specific error. But maybe it's better to just let promscale crash for certain unrecoverable errors

Anything else we need to know?: