WIP: Update project to 0.0.2 using rclcpp::waitset

[carlossvg]

Summary:

• Remove topic statistics support. It's not a real-time safe approach to report statistics.
• Refactor launch files. Separated in different launch files. Now it's possible to run each node and rviz ndividually.
• Process settings command-line options moved to yaml param files. Process settings added as parameters.
• Refactor process settings tool class. No command-line arguments.
• Created pendulum_tests package to implement integrations tests.
• Removed some integration tests. Integration test go now in pendulum_tests.
• Removed pendulum_demo package. Executables are now self-contained in their package. There is no composition executable. Each node runs in its own process.
• Removed deadline QoS support. Deadlines are detected using wait-sets.
• Created real-time loop using a wait-set.
• Launch files run nodes in separate processes.
• Real-time tasks are run in a separate thread.
• Shared variables between real-time thread and non-real-time thread use RealtimeObject or atomics.
• Add test to catch dynamic memory allocations in real-time safe functions.
• Add common PendulumStatus data struct.
• QoS depth changed to 1. We only care about most recent message.
• Default update period changed to 1ms.

TODO:

• [ ] Bump version to 0.0.2
• [ ] Update instructions
• [x] Document code
• [x] Add RT params
• [ ] Add duration option

[ZhenshengLee]

Greate to see that waitset being used in ros2 apps, which is really rare in community. The only material about waitset in ros2 is the ros2 examples and the PPT from Apex.AI in ROSCon.

Regarding to this pr, What's the current status? Can this pr be devided into multiple parts to get merged step by step?

EDIT: autoware.universe is migrating it's ros2 apps to waitset-based subscription. see https://autowarefoundation.github.io/autoware-documentation/main/contributing/coding-guidelines/ros-nodes/topic-message-handling/

and more examples about waitset subscription https://github.com/takam5f2/ros2_subscription_examples

[carlossvg]

I think it makes sense to split this PR into smaller parts. If the waitset change is interesting we could merge that in a separate PR.