Inlining and customizing sampling functions to reduce memory traffic and compute
Improve calcOrientation implementation.
Using more efficient rounding routines.
Removing unnecessary use of local memory

Added the defaultNorm() method to the DescriptorExtractor class. This method returns the default norm type for each descriptor type. The tests and C/C++ samples were updated to get the norm type directly from the DescriptorExtractor inherited classes.

This was reported in feature report #2182 (http://code.opencv.org/issues/2182). It will make it possible to get the norm type usually applied matching method for each descriptor, instead of passing it manually.

Conflicts:
	modules/core/include/opencv2/core/cuda.hpp
	modules/cudacodec/src/thread.cpp
	modules/cudacodec/src/thread.hpp
	modules/superres/perf/perf_superres.cpp
	modules/superres/src/btv_l1_cuda.cpp
	modules/superres/src/optical_flow.cpp
	modules/videostab/src/global_motion.cpp
	modules/videostab/src/inpainting.cpp
	samples/cpp/stitching_detailed.cpp
	samples/cpp/videostab.cpp
	samples/gpu/stereo_multi.cpp

Also changed the static functions to inline, to stop GCC complaining
about unused functions.

cudev is used for new device layer

it uses HAVE_CUDA, HAVE_NVCUVID, etc. macroses

Looks like we don't actually use them; when creating precompiled headers
with Visual C++, a dummy .cpp is created at build time.

ReST directive was used. Also fixed some other ReST directives that were
not correct and removed some warnings during buildbot checks.

- Used lunix style slashes - works on most systems
- Removed all trailing whitespaces

Turns out, you can't use preprocessor directives inside macro arguments.
Who'd have thought?

Now it allows choosing between arbitrary implementation variants.