[kuka_ros2_control_support] Fix ros2 control support test

[muritane]

without pre-commit currently

[gwalck]

I like the revised access to the possible test choices, using python API instead of subprocess.

The tests pass, but the new joint_trajectory test should be slightly improved. I did not manage to make the test fail. I tried setting joints in URDF with lower=upper but the mock component does not respect joint limits, so not possible to see if a robot has bad limits that way.
What would be an example of bad URDF that would make the trajectory test fail ? (to see if the test can actually fail)
A bad topic name on the trajectory controller makes the wait function thrown an exception and not fail properly.

After some discussion with @destogl one should ensure that a broken setup between the kuka robot description and the ros2_controller additional description is captured by the test. Currently if a joint name is different than joint_a# in the robot URDF, the tests still pass but the robot_state_publisher does not receive correct joint_names and hence the robot is all white in RViz. Even if we set the command_names for the controllers to random names, the tests pass, because the mock interface uses those and the test relies on matching output/input of the controller_state, not on actual robot joint_states matching controller_state names.

At least wrong names in the additional ros2_control URDF and non-matching names in their param config fails.

[gwalck]

@muritane you wrote "without pre-commit" currently. Do you plan to fix the pre-commit issues ? Other than that, I am happy with the changes. Thanks.

[destogl]

If you are fixing pre-commit, please also update this option: StoglRobotics/ros_team_workspace#135

[gwalck]

@muritane , could you also make use_mock_hardware to true by default in the test_bringup.lauch.py

kuka_experimental/kuka_ros2_control_support/launch/test_bringup.launch.py

Line 154 in c4f2d31

default_value="false",

because current set of default values leads the manually started test_bringup to crash (RSI use_rsi_communication and EKI use_eki_communication are both to false as well leaving not hardware_interface instantiated at all) even with the minimum mandatory arguments given.

[muritane]

for fixing precommit there is separate pr and still wip: #30