Add ROS executor queue watchdog

[jon-mcmillan]

Summary

There is a high likelihood that in some cases the ROS executor queue stops making progress. This adds a watchdog that probes queue progress and forces a restart when the queue stays blocked past the recovery timeout.

Changes

• Add a ROS executor watchdog that submits periodic no-op probes through the executor queue.
• Reuse the existing health.watchdog config for both LiveKit recovery and ROS executor progress.
• Update package version to 0.1.2.

Notes

The watchdog runs outside the ROS executor, so it can still detect when executor-backed work stops moving:

every 5 seconds:
  probe = ros_executor_queue.submit(no_op)

  while probe is not finished:
    if probe has been pending longer than recovery_timeout:
      exit process so the container can restart the bridge

    if probe has been pending longer than 2 seconds:
      log executor unhealthy

    sleep 100 ms