Added sphere-capsule routines

Memory leak in BVHDistance

No commit message

Fix template specialization problem that caused link errors in Visual Studio 2010

fix the crash caused by the typo in intersect_Triangle

change the normal direction returned for triangle-triangle intersection, now it is from object 1 to object 2, consistent with shape intersection returned by libccd and gjk.

git-svn-id: https://kforge.ros.org/fcl/fcl_ros@197 253336fb-580f-4252-a368-f3cef5a2a82b

git-svn-id: https://kforge.ros.org/fcl/fcl_ros@196 253336fb-580f-4252-a368-f3cef5a2a82b

git-svn-id: https://kforge.ros.org/fcl/fcl_ros@195 253336fb-580f-4252-a368-f3cef5a2a82b

- In intersect.h/cpp: Replaced some STL functions with their boost counterparts because VS 2010 does not fully support C99/C++11.
- In test/test_fcl_utility.h: Added NOMINMAX definition in order to avoid name clashes on Windows.
- In test/test_fcl_math.cpp: Only include simd/math_simd_details.h if FCL_HAVE_SSE is set. Caused compilation errors on Windows because simd/math_simd_details.h includes GCC-specific instructions.
- In test/test_fcl_math.cpp: Converted the constant parameter of the sqrt function from int to double because Visual Studio was not able to deduct the correct version of the overloaded sqrt function.
- In narrowphase/narrowphase.h: Fixed some issues with template specialization. It looks like some parts of the code were not correctly replaced after some copy-and-paste. This caused specialized templated member functions to not be declared in the header file. Thus, Visual Studio used the generic version instead. But, since the cpp defines specialized versions, the linker complained about functions being defined twice. Don't know why it compiled with GCC.
- In BVH/BV_splitter.cpp: It seems like Visual Studio does not automatically compile a fully specialized version of a templated class even if a specialized member function is defined. Thus, needed to add statements to force a compilation of specialized class templates.
- In BVH/BVH_model.cpp: Fixed a bug that caused a segmentation fault somewhere else (probably platform-independent).
- In ccd/taylor_model.cpp: Fixed a typo that did not cause any trouble for me but might when checking assertions.

git-svn-id: https://kforge.ros.org/fcl/fcl_ros@194 253336fb-580f-4252-a368-f3cef5a2a82b

git-svn-id: https://kforge.ros.org/fcl/fcl_ros@193 253336fb-580f-4252-a368-f3cef5a2a82b

git-svn-id: https://kforge.ros.org/fcl/fcl_ros@192 253336fb-580f-4252-a368-f3cef5a2a82b