Robin Leadbeater

That does look rather like pincushion distortion 


(In stellar spectra, atmospheric refraction can also bend the spectrum but this cannot be the cause here)

Lens distortions will also distort the spectrum along the dispersion direction as well as bending the spectrum. The DSLR lens is probably better in this respect than the CCTV lens. What does the image look like without the grating eg   eg does the roof line look curved as you move from top to bottom of the frame ? 

If the lens is distorting the image significantly that will make precise calibration difficult. Once defined you could correct the distortion first in software first which should straighten everything out and linearise the dispersion. I don’t have any direct experience of handling this with meteors and wide angle lenses though. Perhaps Bill Ward has come across it and has some ideas on how to deal with it  ?