TODO file

Bug at the dumping of coefficients (all nodes dump only at the first instance)
Since the jerk calculate from the SCF is zero initially, the zeroth timestep for each particle is calculated through a particular if-clause in phiGRAPE and the result is that it is quite large (at least in the purely collisionless case).
Fix CEP so it doesn't need a preprocessor definition.
Read DTSCF from a file or something.
Memory_buffer class to use pinned memory. Also add a proper destructor.
Add a print statement with the content summary of the grapite.mask file.
Make it so that very hard binaries merge if there is zero softening.