updated ORB: limit the number of output keypoints, use bi-linear interpolation between subsequent layers instead of much slower area-based interpolation between 0-th and i-th layers.

[*] Fixed two bugs in reduction functor: out of shared memory bounds access and missing volatile on GF100 and further

Optimized version of 32F Norm function demands bigger MaxErr value, because it calculates norm in floats instead of double.

Fixed small problem in chart.py.

Changed the class DetectionBasedTracker -- changed the system of states in the subclass DetectionBasedTracker::SeparateDetectionWork, made them more logical.

Android cmake toolchain: changed some error messages, fixed error with NDK r5, improved includes and link directories configuration