Stop using half-steps for A and B

[egavazzi]

Fixes #36

We were using half-steps in height for the A and B matrices. The problem came from doing this with the A matrix but not with the Q matrix. This is (temporarly?) fixed by removing the half-steps for calculating A and B.

Explanation

To explain the problem lets take an example:
For a stream of electrons coming down, at an altitude z and energy E we were 'removing' electrons degrading in energy or scattering in pitch-angle. This was done using half-step in height so using neutral densities at altitude (z + 1/2). But now when calculating the matrix Q the 'source' of electrons at lower energies, we are not using neutral densities at altitude z, without taking half-steps. As neutral densities are higher at lower altitudes, this led to an imbalance in the production and loss of electrons, with more electrons being produced than removed.

Below are profiles of Ie at different energies and varying dz.
With halfsteps:

Without halfsteps:

(x-axis Ie, y-axis altitude)