Test cases (accuracy and performance) are provided.

When argument pointers pushed into an vector and the pointers point to
address on stack, we need to make sure they are valid until kernels are
successfully flushed onto the queue.

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.

And a potential problem in remap

remove as much warnings as possible
use enum instead of MACRO for ocl.hpp
add command line parser in accuracy test and perf test
some bug fix for arthim functions

change the download channels to oclchannles()
fix bugs of arithm functions
perf fix of bilateral
bug fix of split test case
add build_warps functions

from my svn 855

fix the channel 3 bug in matrix operation
perf and buf fix for LUT haardetect convertC3C4 resize warpaffine copytom settom
add convovle
remove stereo

add remap

add OCL 1.2 feature for setTo
bug fix of integral
replace the error code -217 with suitable MACRO
simplify tests, no need apply a new context for each test case
add more control for tests in utility.hpp

add channel 3 support
add fast way Between CPU and GPU for the data which is aligned