fix(worker): respect logging configuration on reload

[zhechen-cs]

Description

With --reload and --no-configure-logging, the reload warning calls logging.warning(), which installs a default root handler when none exists. The process manager's logging.debug() call does the same when handling a shutdown or reload event. This can cause duplicate output when the application subsequently configures its own logging.

Use the existing module loggers for both records so worker management respects configure_logging=False. Add subprocess regression coverage for all four combinations of reload and configure_logging, including processing a shutdown event.

Related to #472.

Validation:

• Python 3.12: 375 tests passed.
• The subprocess regression fails on unmodified master and passes with the fix.
• All repository pre-commit checks passed (AST, whitespace, TOML, typos, secret detection, Ruff, Black, MyPy).