Remove lifecycle node controllers

[v-lopez]

Summary of changes:

• Replace LifecyclePublishers with regular Publishers, unfortunately now controllers could publish while deactivated.
• Refactor the calls to get_lifecycle_node

The LifecyclePublisher above made me notice that there are no LifecycleSubscribers or LifecycleServices. Is it because they have yet to be implemented, or because a LifecycleNode stops serving callbacks if it's not active? I could not find mention to this.

[codecov-io]

Codecov Report

Merging #124 (d7c87cb) into master (38f793d) will decrease coverage by 0.00%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master     #124      +/-   ##
==========================================
- Coverage   37.63%   37.63%   -0.01%     
==========================================
  Files          30       15      -15     
  Lines        3762     1868    -1894     
  Branches     2354     1171    -1183     
==========================================
- Hits         1416      703     -713     
+ Misses        246      119     -127     
+ Partials     2100     1046    -1054     

Flag	Coverage Δ
unittests	37.63% <ø> (-0.01%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
...ller/test/test_load_forward_command_controller.cpp
...mand_controller/src/forward_command_controller.cpp
...ler/test/test_load_joint_trajectory_controller.cpp
...te_controller/test/test_joint_state_controller.cpp
...ectory_controller/test/test_trajectory_actions.cpp
...ory_controller/test/test_trajectory_controller.cpp
...ntroller/test/test_load_joint_state_controller.cpp
...include/joint_trajectory_controller/trajectory.hpp
...include/joint_trajectory_controller/trajectory.hpp
...nt_state_controller/src/joint_state_controller.cpp
... and 35 more

[destogl]

All good!

[destogl]

Also remove this line please.

[v-lopez]

Done

[bmagyar]

Ok I've touched things up a bit, added back the node nullptr check to facilitate one of the existing tests and modified another test to not try the ill-posed update() call. I think we should discuss what exactly the role of each state is, where we want to allow memory allocation and where is shouldn't be. When we mixed lifecycle nodes and the back-and-forth mechanism of "let me declare what I want then you assign" with controllers and CM made all of that a little fuzzy, the tests will probably need to be revised on a couple of places to reflect this properly. Not a blocker at the moment though.