* Introducing campvis::Tensor2<T> type
 * Updated MhdImageReader to support TensorImages

* fixed VolumeExplorer/VolumeRenderer not correctly updating the proxy geometry due to missing locks and thus skipped invalidations
* adjusted VolumeExplorerDemo image URL

Conflicts:
	modules/io/processors/abstractimagereader.cpp
	modules/io/processors/abstractimagereader.h
	modules/io/processors/csvdimagereader.cpp
	modules/io/processors/csvdimagereader.h
	modules/io/processors/genericimagereader.cpp
	modules/io/processors/genericimagereader.h
	modules/io/processors/ltfimagereader.cpp
	modules/io/processors/ltfimagereader.h
	modules/io/processors/mhdimagereader.cpp
	modules/io/processors/mhdimagereader.h
	modules/io/processors/rawimagereader.cpp
	modules/io/processors/rawimagereader.h
	modules/io/processors/vtkimagereader.cpp
	modules/io/processors/vtkimagereader.h

Loading can be done with genericImageReader, however, assertion fails for sliders. thread ownership issues

*IMPORTANT FIX* Fixes possible crashes/segfaults when converting OpenGL image representations to local image representations:
 * ImageRepresentationLocal::tryConvert() does not deadlock anymore when called from OpenGL context and converting from GL representation
 * Fixed possible double deletion after converting from OpenGL to local representation due to ambiguous pointer ownership
 * fixed possible out of bounds array write in tgt::Texture::downloadTextureToBuffer()

fixed PropertyCollectionWidget changing visibility of property widgets from non-Qt thread (now using Qt signals).

Loading can be done with genericImageReader, however, assertion fails for sliders. thread ownership issues

Added smallHeart.mhd sample volume to repository, all demo pipelines in vis module refer to this file. Hence, all vis pipelines should work out of the box from now.

Closes Issue #16: Geometry TF editor thread-safety: Reviewed the code + extensive stress testing - fixed a recursive mutex acquisition.

* Moving to Apache 2.0 license
* Updated AUTHORS.txt

* Split up ProcessorDecoratorShading in ProcessorDecoratorShading inheriting from ProcessorDecoratorGradient
* Introducing GlGradientVolumeGenerator processor in preprocessing module

removed getElementXD() and getElementXDNormalized() from core GLSL headers since the same functionality is directly provided by GLSL with texture() and texelFetch() in a much cleaner way.

 * use texture and textureOffset instead of clumsy getElement3DNormalized
 * introducing Sobel filtered gradients

* Updated DepthDarkening to use GlReduction to compute min/max depth

 * Added a stripped version of NLOPT sources as external library, so that the registration module can be built right away.
 * Some minor improvements in CMake files

* Introducing NloptRegistration pipeline performing the optimization
* SimilarityMeasure only computes the similarity and difference image