Add an exit code signaling permanent failure to avoid respawn cycles.

[rsmmr]

Recording an idea here that came up discussing the management framework: Our supervisor currently always restarts Zeek processes after their termination, independent of why they quit. While that's desired for unexpected terminations, like crashes, it's not helpful if Zeek runs into some kind of configuration issue that will need human intervention to solve.

I'm thinking we could use a magic exit code to flag permanent errors for which the Supervisor should not attempt a Zeek restart until some kind of manual action has been taken. Examples would be script parsing errors, trouble with loading plugins, etc. Maybe even any errors that happen before we get to zeek_init()?