GitLab

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

New Fila : Build a blocked tree from either a classic FOctree by copying each cells or from a classic FOctree but reading only the leaf

Fixed FinterpMatrixKernel by moving declaration of constant array member to .cpp file + fixed Chebyshev non homogeneous applyC.

Implemented dense tensorial Chebyshev kernel + add test, fixed some failing utests by comparing RELATIVE errors instead of absolute.

Fixed handling of non homogeneous kernels of interactions (aka matrix kernel) in Uniform kernel (TODO Fix in Chebyshev if relevant), Introduced blockwise evaluation of tensorial matrix kernel (+redesign precomputation), added input parameter in AbstractUnifKernel (for OptiDis non singular Ra_{ijk} or any matrix kernel requiring an extra parameter), +divers test fixing.

Introduced tensorial physical value, potential and force (through templatization of P2PParticleContainer with dimensions NRHS and NLHS, default NRHS=NLHS=1). ChebTensorial kernel does not support this yet (TODO apply same modif on ChebInterpolator).