Binning in LOS projection step should be in pixel space

[wtbarnes]

Currently, when projecting coordinates down to the detector, the binning of the emission is done in world coordinate space using the helioprojective coordinates. Instead, the binning should just be done in pixel space.

This will require refactoring some of the logic around how the header is computed. Currently, the number of bins and bin range are computed in world space using the coordinates of all of the loops and then this is used to compute the header. Instead, the extent of the image would need to be worked out separately, then the header constructed from that estimated extent. A WCS could then be derived from that header from which the pixel coordinates could be worked out. This would also simplify the calling sequence for LOS projection function.