Consistent CI failure for test_intra on build_enabled job

This is just to track the failure.

For example: https://gitlab.com/micro-ROS/ros_tracing/ros2_tracing/-/jobs/312431580

I can reproduce locally using a gitlab-runner.

I think the problem is Fast-RTPS. The ci_source_deps.repos file in ci_base points to master. Changing it to point to 9d562024886f4e1e7be363356ab032ecba934490 like the ros2.repos on master currently does, rebuilding the docker images, and then re-running the job works. I think we could just wait a bit for a fix.