[WIP] World Model refactor and new features

[orduno]

---

Basic Info

Info
Ticket(s) this addresses	#379
Primary OS tested on	Ubuntu 16.04
Robotic platform tested on	Gazeobo simulation of turtlebot

---

Description of contribution

• Refactored nav2_world_model. Defined an interface for specifying different world representations (WorldRepresentation) and moved current functionality to a CostmapRepresentation.
• Added service for checking if a region is free of obstacles.
• Added a RegionVisualizer for displaying regions in RVIZ that get processed by the world model.
• Defined a World Model client class.
• Modified the Global Planner Navfn and the Backup motion primitive to use the new world model client interface.
• Added a new method in the MotionPrimitive class called pathIsClear() that motion primitives can implement for checking if there is enough space for execution.
• Modified Backup so it checks if there is enough space for backing up before executing.
• Added a new nav2 service description for specifying a rectangular region.
• Added a robot_centric_costmap with only the obstacle layer enabled and centered on the robot.

---

Future work

• Obtain robot footprint on motion primitives Obtain robot footprint on motion primitives #524 *

[ghost]

@orduno Can you fix the linter failures before we start reviewing?

[ghost]

@orduno Two things:

1. Smaller PRs please, or make the commit history clean and show the logically separate changes.
2. Can you put together a doc or presentation on the problem you are trying to solve, and the design you'd like to see? It seems like it is something that would be helpful to discuss in the working group meeting.

[ghost]

Comments so far. Still working on it.

[orduno]

@crdelsey

Can you put together a doc or presentation on the problem you are trying to solve, and the design you'd like to see? It seems like it is something that would be helpful to discuss in the working group meeting.

I can put together some slides based on the discussion we had and share it on the working group.

[orduno]

@crdelsey @SteveMacenski I think I've addressed all your feedback. Comments that have been resolved should be hidden, the ones still showing might require your response.

This PR will get pushed to the world_model_refactor branch. There we can work on addressing #517 before merging to master, this is mainly needed because auto-localization need to check for open space even when the robot is not localized.

If everything looks good please approve so I can proceed with merging.

[orduno]

Closing the PR as the main features provided have been incorporated (for example, collision checking, odom/robot centric costmap, etc.) we can continue our design discussion on #565.