fix(cli): add missing subprocess.run() timeouts in gateway

[dlkakbs]

Summary

All subprocess.run() calls in hermes_cli/gateway.py lacked a timeout parameter. If systemctl, launchctl, loginctl, wmic, or ps blocks — e.g. D-Bus socket unavailable, WMI service stuck on Windows, or launchd momentarily unresponsive — hermes gateway start/stop/restart/status/install/uninstall hangs indefinitely with no feedback or escape other than Ctrl+C.

Same class of bug as #3469 (context_references subprocess calls) and #3693 doctor/status subprocess calls). gateway.py was not covered by either.

Root cause

subprocess.run() defaults to timeout=None — waits forever. The rest of the CLI already uses timeouts consistently (clipboard.py, banner.py, doctor.py, status.py). gateway.py was the remaining gap.

Fix

Timeout values applied across all 32 call sites:

Call type	timeout
Lifecycle: start, stop, restart, enable, disable, daemon-reload, launchctl load/unload	30s
Query: is-active, systemctl status, launchctl list, loginctl show-user, journalctl, tail, ps aux, wmic	5–10s
loginctl enable-linger	10s

For _is_service_running() and launchd_status(), TimeoutExpired is caught explicitly and treated as not-running — consistent with how non-zero return codes are already handled in those functions.

All other call sites are either already inside try/except Exception blocks (find_gateway_pids, get_systemd_linger_status, _enable_systemd_linger) or propagate TimeoutExpired as a clear exception instead of hanging forever.

Verification

AST-based check confirms zero remaining subprocess.run() calls without
timeout= in gateway.py:

python3 -c "
import ast, sys
source = open('hermes_cli/gateway.py').read()
tree = ast.parse(source)
missing = [n.lineno for n in ast.walk(tree)
           if isinstance(n, ast.Call)
           and isinstance(n.func, ast.Attribute)
           and n.func.attr == 'run'
           and isinstance(n.func.value, ast.Name)
           and n.func.value.id == 'subprocess'
           and not any(kw.arg == 'timeout' for kw in n.keywords)]
assert not missing, f'missing timeout at lines: {missing}'
print('ok')
"

[teknium1]

Merged via #5424. Your approach — tiered timeouts across all subprocess.run() calls in gateway.py — was the right one. We reimplemented on current main since the launchctl section was rewritten (bootout/bootstrap/kickstart replaced load/unload/start/stop), and added a third tier (90s) for stop/restart commands to respect our TimeoutStopSec=60 in the systemd unit. Thanks for the contribution, @dlkakbs!