motionAnalyzer refactored to account for different exercises

[vvasco]

This PR refactors motionAnalyzer to deal with different kinds of exercises. Specifically:

• We can now deal with:

  • rehabilitation exercises (abduction, internal/external rotation and reaching) and test exercises (timed up and go).

• I introduced the new abstract class Exercise, from which the following classes are inherited:

  • RangeOfMotion for abduction, internal rotation and external rotation;
  • Reaching;
  • Tug. Each exercise has its own metrics to be evaluated and its own feedback, which are defined in the .ini files of the exercise in the section [metrics] and [feedback] (for now no feedback is defined for the TUG). The motion repertoire includes all the .ini files.

• For each metric we can compute several sub-properties and we can select which one to visualize on yarpscope at runtime, through the provided API:

  • listMetricProps: for listing the available properties of the analyzed metric;
  • selectMetricProp: for selecting the property to visualize.

• The class Metric is extended to include the new class Step, for which the following sub-properties can be computed:

  • step length, step width, cadence and speed.

• I updated the offline report accordingly.

• I successfully tested offline the new framework with the Y2M2 demo. To be tested carefully on the robot.

• Doc updated.

[vvasco]

Successfully tested Y2M2 demo on the robot.