It doesn't make sense to export template classes, since there's no way
the user can learn that it can import them. In fact, it is somewhat
deleterious, because every module will export every instantiation
that it uses, but doesn't inline.

3.0 seems like an appropriate time to do that.

Conflicts:
	modules/core/include/opencv2/core/types_c.h
	modules/gpu/src/cuda/imgproc.cu
	modules/gpu/src/cuda/safe_call.hpp
	modules/gpu/src/error.cpp
	modules/gpu/src/imgproc.cpp
	modules/imgproc/src/distransform.cpp
	modules/imgproc/src/shapedescr.cpp
	modules/python/src2/cv2.cpp
	modules/python/src2/cv2.cv.hpp

Update on the class to reflect the review. Split the class into virtual and implementation. change of name to LineSegmentDetector, using Input/Output-Arrays, general clean ups.

Neumann L., Matas J.: Real-Time Scene Text Localization and Recognition, CVPR 2012.

High-level C++ interface and implementation of algorithm is in the objdetect module.
C++ example, a test image, and the default classifiers in xml files.