Handle shutdown state during supervisorctl preflight

[tianrking]

Summary

• handle SHUTDOWN_STATE returned by the shared supervisorctl preflight
• show an actionable retry message instead of the generic XML-RPC exception and source location
• preserve the generic failure exit status for scripts

Fixes #48

Root cause

Every supervisorctl command calls Controller.upcheck() before performing its action. The preflight recognized UNKNOWN_METHOD, but re-raised SHUTDOWN_STATE, so onecmd() formatted the known server lifecycle state as an internal exception with a Python file and line number. This is especially visible when a command follows reload before the server has finished restarting.

Tests

• python -m pytest supervisor/tests/test_supervisorctl.py -q (199 passed on Linux)
• python -m compileall -q supervisor
• git diff --check