Fixed and reformatted setCompressedRowLengths test function. Found error for setRow test function, it is possibly in the implementation of SlicedEllpack.

Added Ellpack and SlicedEllpack to tests. Changed setCompressedRowLengths test to make it work with all Ellpacks. setRow test is broken for SlicedEllpack. Commiting for backup purposes.

Added Type instead of all int indexes. Formatted includes to top of file. Removed the need for using namespaces.

Added Type instead of all int indexes. Formatted includes to top of file. Removed the need for using namespaces.

Added template testing for all working tests. Implemented RealTypes and occasionally IndexType instead of int/float/etc. Updated TODO.

	Added getType test error location, commented out its tests.
	Added visual representation of used matrices in tests.
	Added working and correct version of addElement test.
	Fixed vectorProduct test for Host, CUDA throws illegal memory
access.
	Fixed performSORIteration for Host, CUDA throws segmentation
fault.
	Fixed load/save test to not use a path.
	Added test for print function.
	Reformatted code.
	Added messages thrown for CUDA errors into tests, so that the
whole test doesn't crash after those errors.
	Updated TODO.

Added another approach using Vector.h for test of vectorProduct function. Added a concept test of performSORIteration function. Added test for save and load functions. Updated TODO.

Adding commented-out, provisional and non-functioning test for vectorProduct function. Adding only for backup purposes.

Added format dependent test for the setElement function. This is format dependent, because setElement won't work without setting rowLengths.

	Added TODOs to implement rowPointer test for
setCompressedRowLengths and setDimensions test functions.

assuming that getRowLength() works correctly and doesn't need to be
tested.

Added test for setDimensions function. This test is correct assuming that getRows() and getColumns() work correctly and don't need to be tested.

Divided testGetType into separate functions, one for Host and one for CUDA, to avoid non-CUDA systems having difficulties.

	Was using EXPECT_STREQ, instead of EXPECT_EQ.