About the placement of interference filters behind the eyepiece... Mike is correct that the issue revolves around the AFoV. If the NVD lens was of a shorter f.l. and hence supplied a wider FoV, then in conjunction with a wide AFoV eyepiece bandpass de-tuning in the outer field would become worse.
If the NVD lens is of some limited FoV, then it will be the limiter, providing the same angular field limit irrespective of the eyepiece's AFoV (unless the latter is smaller than the former.)
Back in my film days I tried a UHC filter on a 50mm f.l. camera lens, which has a FoV roughly like that of a typical eyepiece AFoV. The outer field color shift was ugly, and response to nebulosity was notably impaired.
And so an interference filter of similar or narrower bandpass tasked with covering a similar field angle via placement between eyepiece and NVD lens will suffer de-tuning in the outer field.
About the use of a focal reducer with an eyepiece...
If the reducer results in an exit pupil larger than the NVD lens's entrance pupil, the image surface brightness does not increase from that obtained when the pupils match. You only get a reduced image scale (with concomitantly reduced resolution) and fainter stars captured.
About image scale and brightness...
Even though a smaller image scale resulting from a longer f.l. eyepiece might provide higher image surface brightness, the poorer resolving power often effectively fails to realize the impression of an improvement because finer, higher brightness structures are blended with dimmer bits so as to yield an averaged, blended object. That is, a dimmer but larger image can be more impressive simply due to the revealing of the smaller details.
The better way to quantify image brightness gains is to use a field-filling uniform-brightness source, such as a dimly lit wall indoors. No need to worry about focus whatsoever, for a uniform flux by definition is unfocused. Such a testing protocol will not be prone to such effects as point source depth of penetration and resolution of object details, both of which profoundly modify impressions of performance.