Implement a Kalman filter for sensor fusion

[HipsterSloth]

Right now we are using a simple LowPass filter on the position data we get from the tracker. On psmoveapi I made a simple Kalman filter for smoothing, but it didn't leverage the accelerometer so it felt laggy:

https://github.com/HipsterSloth/psmoveapi/blob/master/src/tracker/psmove_kalman_filter.cpp

We already have a framework in place for filtering here:

https://github.com/cboulay/PSMoveService/blob/master/src/psmoveservice/Filter/PositionFilter.h

We need to create a new kalman filter that plugs into our filter framework and leverages the other sensor data to make a better prediction of position.

Other helpful links: Pictorial of kalman filters- http://www.bzarg.com/p/how-a-kalman-filter-works-in-pictures/ Tutorial of extended kalman filters (with TinyEKF implementation)- http://home.wlu.edu/~levys/kalman_tutorial/

[HipsterSloth]

Closing this issue as a dupe of Issue #13