If we have perfect matches (min_dist == 0.0), then strict comparison
fails. Making it non-strict results in treating perfect matches as
good.

replaced wrong check to correct check

* removed OpenGLFuncTab, now extensions are loaded internally
* added support of GlBuffer and GlTexture2D to InputArray/OutputArray
* added ELEMENT_ARRAY_BUFFER and PIXEL_PACK_BUFFER targets
* added copyFrom/copyTo method for GlBuffer and GlTexture2D
* removed GlFont
* removed pointCloudShow
* removed OpenGLCleanCallback
* added Access parameter to GlBuffer::mapHost
* added autoRelease parameter to all create methods

* added DenseOpticalFlow interface
* moved OpticalFlowDual_TVL1 to src folder

modified SIFT to 1) double image before finding keypoints, 2) use floating-point internally instead of 16-bit integers, 3) set the keypoint response to the abs(interpolated_DoG_value). step 1) increases the number of detected keypoints significantly and together with 2) and 3) it improves some detection benchmarks. On the other hand, the stability of the small keypoints is lower, so the rotation and scale invariance tests now struggle a bit. In 2.5 need to make this feature optional and add some more intelligence to the algorithm.

added test that finds a planar object using SIFT.

These codes should be included into regular builds.

This reverts commit 54a202b3.

Conflicts:
	modules/core/doc/command_line_parser.rst
	modules/core/include/opencv2/core/core.hpp
	modules/core/src/cmdparser.cpp
	modules/gpu/perf/main.cpp
	modules/gpu/test/main.cpp
	modules/ts/src/ts_perf.cpp

expanded cv::threshold parallelization to other threading frameworks; fixed potential bug with unprocessed bottom of the image; fixed build problem in stitching