Added draco node-level transport pub/sub

[ahcorde]

Added draco node-level transport pub/sub

Claude Opus 4.7

[mergify]

Tick the box to add this pull request to the merge queue (same as @mergifyio queue).

• Queue this pull request

[asymingt]

Don't you have have to add nodes to the executor before you use them?

exec.add_node(pub) 
exec.add_node(sub) 

[asymingt]

I'm not a huge fan of timeouts being used in this way. I'd much prefer this to be written as a launch_pytest test using lifecycle nodes, where the test launches and then waits on the state transition of the subscriber to move to READY before firing a timer to send out point clouds at a specific rate. The 15s should be specified deadline specified like this ament_add_ros_isolated_gtest(my_isolated_test test/test_my_node.cpp TIMEOUT 15). Why:

• consolidating timeouts in the build infrastructure is cleaner than ad-hoc within tests
• lifecycle nodes let you observe actual state of a subscriber before sending messages to it
• the subscriber can shutdown on reception of the first cloud, apply the check and convey it within a return code that gets communicated to the launch_pytest and use to pass/fail the test.

At a minimum we need to move the test timeout to CMakeLists and use ament_add_ros_isolated_gtest to mitigate destructive interference (@cottsay this sort of PR is what causes nondeterminism in CI, right?).

Ideally, we'd use launch_pytest and lifecycle nodes to make state directly observable in stead of using time as a weak proxy.

[asymingt]

Feel free to push back if you don't agree with my feedback.