GitLab

Some more cleanup in MatrixKernel class (renaming DIM to NCMP, to avoid confusion in other applications) as well as in Chebyshev/Uniform Tensorial tests.

Conflicts:
	Src/Kernels/Uniform/FUnifTensorialM2LHandler.hpp

Redesigned direct DFT and FFT classes (provided a separate complex valued FFT and templatized direct DFT with physical value type); Provided test for C-interfaced Lapack functions such as Cholesky Decomposition (TODO add SVD and QRD).

New balance algorithm for building the parallel tree. A Balance repertory has been created, with an FAbastractBalanceAlgorithm providing what it's needed to be a BalanceAlgorithm

Introduced core width in R_IJ and R_IJK kernels; Fixed R_IJK (scheme: p_i=R_ijk*q_jk, BEWARE: antisymmetry and non_homogeneity strongly affect P2P and MatrixKernel, force not yet implemented); Improved consistency with potential theory for tensorial matrix kernels by defining number of phys vals NPV and number of potential vals NPOT (NRHS=NPV, NLHS=NPOT*NRHS).

Changes to FFmmAlgorithmThreadProc and Spherical Cell in order to compile with Intel without having any problem with the GetSize function