Test cases (accuracy and performance) are provided.

The bug is a buffer overrun when border type is reflect101.
It is found that gfft crashed with input of size 100x100 on Intel CPU.

When user provided corners buffer is big enough to be copied to from
tmpCorners_, we allow the buffer to be reused other than allocate a new
cl_mem object.

For example, haarcascade_frontalface_alt2.xml is now supported.
Note that classifier's pattern of a cascade file must be consistent,
i.e., all trees must either have two nodes or one node, otherwise
unexpected results will occur.

Other fixes:
Test cases are updated.
Some unused codes are removed.
Fix some problems of haar when using OclCascadeClassifierBuf.

This fix is a workaround for current 2.4 branch without introducing an
additional oclMat buffer into CannyBuf object.
Test case is cleaned up.
Volatile keywords in kernels are removed for performance concern.

Additional notes with this commit:
1. Add cornerHarris_dxdy and cornerMinEigenVal_dxdy to get
the interim dx and dy output of Sobel operator;
2. Add minMax_buf to allow user to reuse buffers in minMax;
3. Fix an error when either min or max pointer fed into minMax is NULL;
4. Corner sorter temporarily uses C++ STL's quick sort. A parallel
 selection sort in OpneCL is contained in the implementation but disabled
due to poor performance at the moment.
5. Accuracy test for ocl gfft.

1. Use macro defines for some parameters(radius).
2. Reduce local memory usage.
3. Fix accuracy problem on Intel GPU.

Also cv::, cv::gpu:: and cv::ocl:: namespace prefixes can be safely omitted
inside CV_ENUM and CV_FLAGS

RadiusMatch for HammingDist cannot pass yet.

OpenCL StereoBeliefPropagation, ported from GPU implementation.

The test cases are ported from gpu module (masked version is still
missing).
Key point direction definition is synchronized with cv::SURF.
nonfree module dependency is added into ocl module's cmake file.