Fixed a couple of memory leaks in videoInput::getDevice() when multiple devices are present in the system

- CV_RGB() macro is moved into opencv2/imgproc.hpp from imgproc_c.h
- samples/cpp/filestorage_base64.cpp is dropped

My build fails with the error:
"C:/PROGRA~1/MINGW-~1/X86_64~1.0-P/mingw64/lib/gcc/x86_64-w64-mingw32/7.2.0/include/c++/cmath:1136:11: error: '::hypot' has not been declared".
I have tried to fix it by adding "#include <cmath>" before the line "#include <Python.h>" but then the build has failed with the error:
"C:/PROGRA~1/MINGW-~1/X86_64~1.0-P/mingw64/lib/gcc/x86_64-w64-mingw32/7.2.0/include/c++/math.h:91:12:: error: 'std::_hypot' has not been declared".
Adding "#include <math.h>" allowed me to build opencv.

Sets a higher limit on videoio's AVI container's maximum chunk size, and adds an explanation of the assertion. Closes #11126

Fix batch norm in training phase.

  * disable carotenete when calling resize on ARM/Aarch64
  * loosen eps on Aarch64

  * Stop calling nppSetStream

* fixed bug #11100 Integer overflow in kmeans

* fixed integer overflow in other divUp-s in kmeans code
fixed warning about size_t to double conversion

Apparently the blog has been updated and the URL has changed, it is still there though.

* Fix #9363

* Renamed the structure and added a new function to the LineSegmentDetectorImpl class as a static member

* Added a new function to the LineSegmentDetectorImpl class as a static member