* Add functionality to filter homography decompositions

* documentation + small refactor

* fix comparing int to size_t (compiler warning)

* fix whitespace issues

* clarification of function return values in documentation

* refactor of function parameters and change in loop nesting

* cleanup useless .h, fix size_t to int compare, small refactor

* fix documentation and whitespace

* change output from return value to outputarray parameter

* update function docs to reflect changes in parameters

* whitespace

* failing test

* fixed warnings related to extended initialisers and improper types

* initialize vectors from arrays

* initialize vectors from arrays part 2

* fix whitespace

* fix trailing whitespace

* Include <inttypes.h> in test_filter_homography_decomp.cpp, should fix 'uint8_t' : undeclared identifier error

* updated the test (made it shorter and providing better diagnostic) and significantly improved implementation (get rid of heavy repeated and/or unnecessary operations)

* fixed compile warning; removed trailing whitespace

* StereoBM: fixed SIMD processing for fixed-type output arrays

* changed norm type and threshold, added assertion

* fixed disp_shift

  * add ifdef to apply on specific situation

* make tmpRow proper size before copyTo to avoid reallocated tmpCol

* do the transposition without creating temporary array; replace TAB with spaces.

* revert the previous commit

Found via `codespell`

Fix unstable work of stereoRectify function. Issue description: https://github.com/opencv/opencv/issues/10791

 calib3d: add stereoCalibrateExtended (#10667)

* cvCalibrateCamera2Internal: simplify per view error computation

* calib3d: add stereoCalibrateExtended

- allow CALIB_USE_EXTRINSIC_GUESS
- returns per view errors

* calib3d: add stereoCalibrateExtended test

* Newton's method can be more efficient

when we get the result of function distortPoint with a point (0, 0) and then undistortPoint with  the result, we get the point not (0, 0). and then we discovered that the old method is not convergence sometimes. finally we have gotten the right values by Newton's method.

* modify by advice  Newton's method...#10574

* calib3d(fisheye): fix codestyle, update theta before exit EPS check

This reverts commit 5384a220.

Allow to use 3 points in SolvePnP if SOLVEPNP_ITERATIVE and useExtrinsicGuess==true. Add bibtex citations for P3P. Update SolvPnP tests.

If there are no OpenCL/UMat methods calls from application.

OpenCL subsystem is initialized:
- haveOpenCL() is called from application
- useOpenCL() is called from application
- access to OpenCL allocator: UMat is created (empty UMat is ignored) or UMat <-> Mat conversions are called

Don't call OpenCL functions if OPENCV_OPENCL_RUNTIME=disabled
(independent from OpenCL linkage type)

it will allow to build features2d even if flann module is not available