Improvement of the spline fitting of the master_sensitivity

[gildepaz]

In some VPHs the master sensitivity might drop suddenly to 0 due to the lack of data at the very extremes of the wavelength range. This affects the master sensitivity derived from spline fitting even in adjacent regions where there is actual information (see also #287). I think the sudden drop to 0 could be easily identified (I am working on that) and then the first and last nodes could be set accordingly.