Refactoring the image_pool for android, and adding some common utils for camera configuration.  Also experimenting with optimization - grayscale preview is way faster than color right now.

as per the tickets: #712 and #713

moved some old stuff to the legacy module; merge "compat_c.h" headers and moved to the legacy as well. moved implementation of many non-critical/obsolete inline functions and methods to .cpp to improve Opencv build time

2) fixed gpu bm test fail
3) CMAKE_MINIMUM_REQUIRED(VERSION 2.8.3)

added first version of gpu::countNonZero for all data types, it doesn't support compute capability 1.0 yet, also fixed some little bugs

fixed the ordering of parameters in PyArg_ParseTupleAndKeywords, added correct Ptr<CvDTreeSplit>::delete_obj() (ticket #406)

added tests for some detectors; made features2d object create functions as static classes methods; fixed OpponentColorDescriptorExtractor, BriefDescriptorExtractor (on rgb); renamed DynamicDetector

renamed PooledAllocator::malloc to PooledAllocator::allocateBytes, Matrix::free to Matrix::release to avoid conflicts with VisualStudio