Gemm openmp improve performance

Replacing the ijk gemm algorithm with ikj  gave me a performance gain of 3x. (Size=2000)

See merge request !85

 Changes to be committed:
	modified:   ../CMakeLists.txt

 Changes to be committed:
	modified:   ../CMakeLists.txt
	new file:   CMakeLists.txt
	modified:   README.md

	modified:   src/Makefile
	modified:   src/pddp_2means.c
	modified:   src/pddp_2means.h

 Changes to be committed:
	new file:   README.md

 Changes to be committed:
	modified:   src/pddp_2means.c
	modified:   src/pddp_2means.h

On branch pddp2means
Changes to be committed:
	new file:   README
	new file:   data/40k.csv
	new file:   src/pddp_2means.c
	new file:   src/pddp_2means.h

Added OpenMP code samples for gemm and LU decomposition(Doolittle algorithm)



See merge request !72

added guards and messages to CMakeLists.txt from dense_linear_algebra,monte_carlo_methods,sparse_linear_algebra,structured_grids dwarfs