WinRT parallels library is causing conflicts in VC++ due to inconsistent usage with C++/CX which is now present in parallels.cpp.  This support should be included in all projects with WinRT specific code except if doing special native C++ tasks.

Fix C++/CX support for all WINRT and bring back native-mode only support which was removed

related issues: #5423 #5476

PR #2968: cce2d992 8578f9c5
Fixed bug which caused crash of GPU version of feature matcher in stitcher

The bug caused crash of GPU version of feature matcher in stitcher when
we use ORB features.

PR #3236: 5947519f
Check sure that we're not already below required leaf false alarm rate before continuing to get negative samples.

PR #3190
fix blobdetector

PR #3562 (part): 82bd82e2
TBB updated to 4.3u2. Fix for aarch64 support.

PR #3604 (part): 091c7a38
OpenGL interop sample reworked not ot use cvconfig.h

PR #3792: afdf319f
Add -L for CUDA libs path to pkg-config

Add all dirs from CUDA_LIBS_PATH as -L linker options to
OPENCV_LINKER_LIBS. These will end up in opencv.pc.

PR #3893: 122b9f8e
Turn ocv_convert_to_lib_name into a function

PR #5490: ec5244a7
fixed memory leak in findHomography tests

PR #5491: 0d5b739d
delete video readers

PR #5574

PR #5202

github editor does not show white space and inserts one for you ... great combination

Check for identity resize => use copy instead (1.37 ms -> 0.07 ms on 720p)

Add py_hdr tutorial under py_tutorials\py_photo.
This tutorial is similar to the one which is given at the C++ tutorials.

major time consuming part for training app is in function fillPassedSamples for negatives, 
this change make SetImage quicker, in a test of my own data, the total time for SetImage decrease from 9177666 to 5839263,
only help for Haar feature and non Haar::ALL mode which is the most commonly case

connectedComponents and connectedComponentsWithStats in
imgproc accept 8-bit single channel image only for input.