Descripe the feature and motivation

All imgproc G-API operations can be found there: https://github.com/opencv/opencv/blob/4.x/modules/gapi/include/opencv2/gapi/imgproc.hpp

Most of them already have prefix GAPI_EXPORTS_W that means they're exposed to python, but a few of them still not.
The main obstacle is signature of that functions is complex from python parser/generator perspective.
Need to invent how composite types such as GArray<GArray<Point>> can be handled by parser/gen automatically without hand-written implementation

Missing operations:

• GArray<GArray<Point>> findContours(...) - 2 overloads
• std::tuple<GArray<GArray<Point>>,GArray<Vec4i>> findContoursH(...) - 2 overloads
• GOpaque<Vec4f> fitLine2D(...) - 4 overloads
• GOpaque<Vec6f> fitLine3D(...) - 4 overloads
• GMatP NV12toRGBp(...)
• GMatP NV12toBGRp(...)
• GMatP resizeP(...)

Additional context

No response