Add a watchdog for OpenClaw gateway recovery

[webup]

What

• Start a ClawMaster-managed watchdog in service mode to monitor the OpenClaw gateway and restart it after unexpected downtime.
• Add CLI visibility for the safeguard, including clawmaster serve help/startup output and clawmaster status reporting.
• Surface watchdog state and restart counts in the Gateway UI with localized copy and test coverage.

Why

Users reported that the OpenClaw gateway can crash from time to time. ClawMaster should make the service mode more resilient and make that protection visible so users can tell the gateway is being monitored.

How

• The backend service starts an opt-out watchdog when launched through clawmaster serve, probes gateway health, restarts it after unexpected downtime, and pauses after an intentional manual stop.
• The CLI passes the watchdog env flag to the backend, documents --no-gateway-watchdog, and reads /api/gateway/status for status output.
• The Gateway page renders safeguard availability, health state, and restart count using localized strings.

Testing

• npm run test:cli
• npm run test --workspace=@openclaw-manager/backend -- gatewayService.test.ts
• npm run test --workspace=@openclaw-manager/web -- src/modules/gateway/__tests__/GatewayPage.test.tsx
• npm run test --workspace=@openclaw-manager/web
• npm run build:backend
• npm run build
• git diff --check
• Browser Use walkthrough against http://127.0.0.1:17223/gateway, including diagnostics navigation and recovery after an induced gateway process stop.

Screenshots

Browser Use verified the Gateway UI locally: the page shows 守护: 自动重启已启用, 状态: 健康, and the restart count increasing after a simulated gateway crash. No screenshot asset is committed to the repository.