tackle realizability issue

See merge request !22

# Conflicts:
#   code/src/main.cpp
#   code/src/solvers/mnsolver.cpp

MVP ML_MN solver stands. Weird mem leak in python. Newton an ML adaption not yet compatible. Huge r/w overhead in ML optimizer

Unify the two branches, since they contain the same work

See merge request !23

newton optimizer works. unit tests for it work. Quadrature Gauss legendre must be on full sphere. Header clean up

merged master into MN_Solver. Adapted MN linesource reference solution to new cfl condition. MN Solver unstable for CFL > .3 right now.

removed commented code in MNSolver. renamed testgroup to validation tests. added scattering to MNSolver

fix #57 and #58. GLquad is now correct and tested for SphericalHarmonics. SphericalHarmonics normality unit test implemented. MN_Solver works (unvalidated) (solves #8). Changed tolarance for quadrature rules, due to weird behaviour of MC quadrature. Its now e-12 instead of e-15