* GAPI: document GCall's recursive behavior

* Move documentation outside the class

Improve stitching detailed (#13584)

* Added block size getter/setters

* Added a bunch of new features to the stitching_detailed sample

* Do not required XFEATURES2D for default use

* Add support for akaze features in stitching_detailed

* Improved sample logs

Add Matplotlib Perceptually Uniform Sequential colormaps (viridis, plasma, inferno, magma, cividis, twilight and twilight shifted).

Fix a bug in cv :: merge when array of 3-channel mat is input (#13544)

* Mat merge function bug fix - Bug fix of merge function of 3-channel vector <Mat> of 3 or 4 matrices

* Add Core_merge test for opencv#13544

* fixups

Channels exposure compensators (#13532)

* feed compatible with single channel images

* Simplified BlockGainCompensator::apply

* ChannelsCompensator

* BlocksChannelsCompensator

* Make source level compatibility detector happy

* dnn/Vulkan: fix GPU hang for heavy convolution tasks

Intel i915 driver will declare GPU hang if the compute shader
takes too long to complete. See
https://bugs.freedesktop.org/show_bug.cgi?id=108947 for details.

The idea in this commit is to divide heavy task into several light
ones and run compute shader multiple times to make each run take
short time enough.

TODO: Add more efficient compute shader
Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com>

* dnn/Vulkan: add a more efficient conv shader