Clarify docs on 'mapped parameters' in README

[hello-binit]

This PR rewords the README to clarify that the parameters zeros and dependent_joints don't actually take in dictionaries, but instead follow the <parameter>.<key>:=<value> format. It also points out that ros2 param list won't show these two params, but that they are still supported. This PR closes #65.

If this change to the README looks good, I can copy it to a wiki. Is there a ROS2 equivalent for the joint_state_publisher ROS wiki?

[clalancette]

So this isn't totally true, or at least it is not meant to be true. If you just run:

ros2 run joint_state_publisher joint_state_publisher

it is true that there will be no declared zero.<joint> parameters. But that's because we don't know what the joints might be.

Instead, if you run:

ros2 run joint_state_publisher joint_state_publisher --ros-args -p zeros.joint2:=3.0

Then the zeros.joint2 parameter will indeed show up in ros2 param list /joint_state_publisher.

Theoretically, at the end of configure_robot, we could do a declare_parameter('zeros.<jointname>', 0.0) for every possible joint that we see in the incoming URDF. But that would mean that on a large robot we would have hundreds of parameters, most of which will never be used. I'm not sure if that is worth it or not.

[hello-binit]

Agreed, that phrase wasn't entirely true. To make it unambiguous that these parameters exist, I've added a commit that declares the zeros and dependent_joints parameters as NOT_SET. The node prints warnings when attempting to interact with these parameters directly instead of using <parameter>.<key>. The README removes the not true statement.

This has the benefit of being able to explore the node's parameters using the cli. In the future, these parameters could have descriptions.

I agree that it's likely not worth declaring every parameter from the URDF; the large printout would make it harder to read the rest of the parameters.

If you think this change doesn't make sense, let me know and I can revert it.

[clalancette]

@hello-binit Thanks for the update! The change to the README looks good.

As for the code change, I'm not inclined to add a parameter that is "fake" and doesn't really do anything. So please remove that change for now.

[hello-binit]

Sure, makes sense. It's removed @clalancette.

[clalancette]

Thanks for the PR, merging. I'm also going to forward-port this onto the galactic and ros2 branches.