Improve spectrometer model handling

[mattias42]

At the moment the handling of the maximum possible intensity from a particular spectrometer model is scattered around in different places in NovacProgram, MobileDoas and the NovacPPP and this is highly error prone and a common cause for problems when introducing a new spectrometer model into the system.

First of all create a table with the supported spectrometer models and their assumed maximum intensity (for a single readout), if the spectra are co-added or averaged and the number of pixels on their arrays (known problem with MayaPro here).

Then make sure that the fields describing intensities in the Novac software have a clearly defined unit (counts per readout, total counts or saturation ratios) and that the handling of these fields is uniform and well defined.

[mattias42]

Also, the model is currently not written to the .pak files (no field for this) but instead NovacProgram guesses the model from the serial number of the device...