issue with mu, cosine of zenith, in absorption and radiation calculation from disk

[cosimoNigro]

The range of mu used for integration changes with the position of the blob above the BH (r) according to

 mu_min = 1 / sqrt(1 + (R_out / r)^2)
 mu_max = 1 / sqrt(1 + (R_in / r)^2)

In the absorption though an array of mu calculated at a single r (the one occupied by the blob) is used
https://github.com/cosimoNigro/agnpy/blob/master/agnpy/absorption.py#L97
when integrating from r to infinity for the optical depth, the array of mu has to be recomputed at each value of r and updated.

For the same reason the same problem might occur in the photon density:
https://github.com/cosimoNigro/agnpy/blob/master/agnpy/targets.py#L202

The integration has to be separated somehow, I'll take care of that.

[cosimoNigro]

The integration is now properly performed by changing the extreme of integrations in mu at each distance.
Taken care by PR #46