* integrate HW version and SW version to same function

  * check compiler more strictly
  * use gcc version of fp16 conversion if it's possible (gcc 4.7 and later)
  * use current SW implementation in other cases

  * enable fp16 feature correctly with gcc on x86/x86_64

  * update the comment to real header

  * remove useHW option
  * update test

  * check compiler support
  * check HW support before executing
  * add test doing round trip conversion from / to FP32
  * treat array correctly if size is not multiple of 4
  * add declaration to prevent warning
  * make it possible to enable fp16 on 32bit ARM
  * let the conversion possible on non-supported HW, too.
  * add test using both HW and SW implementation

  * modify test, too

Command line generation routine uses "-isystem" to include headers outside of "<opencv>/modules" folder, but GCC 6 does not work when passed "-isystem /usr/include" option.

  * fix issue 6521
  * use correct comparison

All of these: (performance) Prefer prefix ++/-- operators for non-primitive types.
[modules/calib3d/src/fundam.cpp:1049] -> [modules/calib3d/src/fundam.cpp:1049]: (style) Same expression on both sides of '&&'.

Add documentation for Kernel::run