document procedure to kill stale ros2 daemons

[berndpfrommer]

This is related to issue #184

[Yadunund]

minor nitpicks

[Timple]

Maybe not the correct place to ask. But might ROS be improved such that the daemon is only spawned for rmw's that need it?

[berndpfrommer]

Accepted your suggested changes, LGTM.

[Yadunund]

Maybe not the correct place to ask. But might ROS be improved such that the daemon is only spawned for rmw's that need it?

It may be worth having a discussion on https://github.com/ros2/ros2cli/tree/rolling. But imo even rmw_zenoh benefits from the daemon (which will be started automatically when there isn't one running and your run a ros2 topic/node command) since these graph queries can be made without initializing a new ROS context and hence waiting to connect with the Zenoh router; instead, we can query this information from the daemon node that has already connected to the router. Perhaps ros2 shutdown command will be helpful to ensure all ROS 2 nodes and daemons are terminated before users switch to another RMW.