Closes #1613

What changed

• repair stale systemd and launchd service definitions during hermes gateway install instead of skipping whenever a local file already exists
• reload launchd jobs on hermes gateway start when launchctl start fails because the job is unloaded
• make launchd_restart() tolerate an unloaded job and reuse the new start recovery path
• stop hermes gateway restart from silently falling back to a foreground gateway when a configured service manager restart failed
• extend launchd status so it reports the local plist path, whether the service definition is stale, and that the service is present locally but not loaded
• add regression tests for stale install repair, launchd start recovery, launchd status diagnostics, and restart failure handling

Why

• a local plist/unit file existing is not enough; it can still be stale or unloaded after repo moves, failed loads, or partial service breakage
• the previous foreground fallback hid real background service failures and made recovery look successful when it was not
• users need status output that distinguishes "not installed" from "installed locally but stale/unloaded"

Breaking changes

• None

Testing

• python -m pytest -o addopts='' tests/hermes_cli/test_gateway_service.py tests/hermes_cli/test_update_gateway_restart.py -q
• manual read-only check on macOS: python -m hermes_cli.main gateway status

Platform

• macOS 15.6 / Darwin 24.6.0 x86_64
• Python 3.11.14