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

[~] NCVPyramid uses tr1 and thus can be compiled with CL, commented out on linux
[+] Moved reduction functors to NCVAlg
[*] Warnings in NCV

[+] Color conversions stub via pixel types
[+] Pyramid calculation (required for mipmaps in CUDA 4.1)
[~] Changed C strings to C++ throughout NCV
[~] Fixed a couple of bugs in NCV