Hi Jacob,
I found the problem! The Niton file is highly unusual - it is providing an energy range of 60 keV. And the spectra exceeds the channel count for the available detectors on the market. The odd terminology you find in the file means that this is a mathematical model of a spectra - there has been some photoshop-style post processing of this spectra. So full warning, the data here may not be a perfect representation of the physical spectrum acquired by the instrument.
That said - the reason the deconvolution is not working is because Artax is confused by the range of energy this was taken at. Artax has energy presets for 10, 20, 40, and 80 keV. This spectrum was purportedly taken with a detector capable of reading 60 keV, though the manufacturer (keytech) is only capable of at best 45 keV with a bit of manipulation (heavier photons will blast right through the 450 micrometer silicon detector).
All this is to say, you have done everything you need to do on the file. What is now needed is to tell Artax an energy range. Go to Analyze -> Elements and in the lower left corner select '40 keV' as an energy range for the Artax deconvolution to work with. This is a software feature I never thought would be needed, but glad it is there.
Let me know if you have any questions,