Open & Closed loop options in the joint limits interface

[po1]

In some cases, feedback might not be available for one of the state variables (position, velocity). The joint limits interface has no knowledge of what is available and what is not.

The idea here would be to be able to choose between open and closed loop (default) joint limits enforcers. The closed loop would be slower, but potentially safer.

Issue related to #1

@adolfo-rt @efernandez

[adolfo-rt]

Yes, I'll try to propagate these changes, and reflect them in the tests/doc.

[po1]

After further thinking, and discussion about the diff drive limiters, it appeared that the open-loop joint limits seem much better, and safer. Maybe we should ask around if there is any good reason to have closed-loop limits.