'did' was only really an external thing, so naming it "SFM" is more clear!

Renamed all references to "ibug2did" to "ibug_to_sfm".

Explicit casts and initializations to glm::vec3

The implicit conversions don't seem to work on OS X clang-3.9.1. A bit odd, since it works on travis on Linux with clang-3.9, 4.0 and 5.0.
Anyway, maybe in the future we can change it back if a Mac clang update fixes this.

BFM conversion is now ~200x faster

Added explicit casts to glm::vec3 from glm::vec4 as it was causing
build errors with clang 3.9.1 on OS X.

Changed assignment operation to initialization for ray_origin.

Removed loops from BFM conversion script.

CMake changes (built in windows and linux):
* Include newest eigen3 CMake file
* Make CMake scripts work with OpenCV 3.1

The diff looks really weird because I clang-formatted the whole document. But there should be no changes in extract_texture, except for tabs->spaces.

* Can now return any cv::Mat to Matlab
* Can convert 3 and 4-channel images from Matlab to cv::Mat

1) Fixes for OpenCV > 2.4.3
2) Update to newest eigne3 cmake file (support in environment variables).

This enables defining multiple functions per Matlab mex module

Can return 4-chan uchar matrices (isomaps) back to Matlab, and convert 3-chan uchar matrices (images) from Matlab to C++.

These are required for extract_texture in Matlab.
Maybe the better strategy would be to add a constructor to RenderingParameters and not have all these setters.

* Convert Frustum back and forth
* Convert RenderingParameters back to C++
* Convert glm::mat4x4 back to C++

Includes my multidim-array PR plus a fix for the overload resolution in ::to/at

Change CMAKE_SOURCE_DIR to CMAKE_CURRENT_SOURCE_DIR

This fixes wrong paths in case someone includes eos in CMake with `add_subdirectory`.

Matlab's default precision is double, that's why we only convert MatrixXd for now. Added a note highlighting this.