SmartMove and config fixes

This release fixes a fairly major bug with the way SmartMove worked. It should be much more predictable now.

Add preliminary kinetic support

CoSTAR should now work both on ROS Kinetic and on ROS Indigo. Thanks to @ahundt @bcjiang and @fjonath1 for putting this one together.

This release also fixes some other issues, refactors and improves several CoSTAR components, and updates our development guidelines.

Servo movement patch and installation updates

Video of our robot in action available here.

This release was tested on the robot!

CoSTAR v1

Version of the CoSTAR operating system tested on the Universal Robots UR5 and a simulated version of the KUKA LBR iiwa. Recent UI fixes include:

• removal of memory leaks that caused some crashes
• fixing a bug by which windows could come detached
• general speed improvements

This version includes the full suite of CoSTAR operations developed to date:

• DetectObjects: call to SP_Segmenter to get objects and TF frames
• SmartMove. SmartGrasp, and SmartRelease: high level queries to create more complex programs
• And the full set of motion, grasping, etc.

SmartMove, IIWA, and the Ui

Improvements to:

• SmartMove, SmartGrasp, and SmartRelease
• UI speed and responsiveness
• IIWA support

Note that there's a known bug in the UI which causes some issues when loading/switching between GUI elements quickly. It was introduced when we fixed a major memory leak in the Instructor code.

UR5 configuration improvements

Includes "perfect" smartmoves, better state validity checking, corrected object orientation.

Planning and Perception

Merge pull request #26 from cpaxton/devel

improvements to perception and p[lanning integration

Smart Grasp and Release

Also added new options for instructor and refactor.

CoSTAR SmartMove Refactoring and Instructor Cleanup

Removed some unnecessary files from Instructor; smartmove has been split into a "query" function so we can use it to do more interesting things.

Preliminary DVRK support added

Thanks to Baichuan Jiang. Also added more general support for teaching with an interactive marker (so we do not need to physically interact with a given robot), and wrapped the "endpoint" publication functionality of the CoSTAR Arm component into the code itself to let us do more interesting things with it.