It is not about all-or-nothing, it is about how much of the STAR TREK filmed multiple series and movies can be accepted as accurate for the viewer. If we see a starship class that we do not think belongs, then just disregard the class? In the Star Trek TOS-Remastered they changed things in the original episodes to correct gaffes or simply enhance the content of the episodes. STAR TREK ENCYCLOPEDIA(Michael & Denise Okuda) has reenforced what we have seen in those official filmed productions. The production team were the creators of the end product so how much can a viewer changed what our lying eyes are seeing?
I wasn't talking about entire classes, but about individual instances where the special effect shown onscreen is clearly a money-saving or logistically unavoidable compromise for something where a different class of ship would be more appropriate -- like using stock footage of the Botany Bay for the robot freighter in "The Changeling" or the Rigel VII fortress matte painting for Flint's mansion (both of which were changed in TOS Remastered, note); or using an unaltered Klingon Bird of Prey model to represent a ship that's described in the script as much larger (which was probably an error); or using a Miranda-class miniature to represent the Lantree, which was described in the script as a supply ship with only 26 people aboard (probably because they didn't have a more suitable miniature available, though it seems an Oberth would've worked okay there); or reusing an Alpha-Quadrant starship design in a Voyager episode (because it's cheaper to reuse an existing model than to build a new one -- and this is apparently just as true of digital models as physical ones). Or the example that started this discussion, the use of a stock D-7 Klingon ship in an ENT episode in place of the earlier model that was intended to go there.
Part of interpreting a visual text such as a television series is recognizing that the execution often represents a set of compromises for budgetary or logistical reasons, so that the end result is more an approximation of the intent than an exact achievement of it. In cases like those described above, I think it's better to favor the intent suggested in the script than to take the images too literally.
Thanks for clarifying, yes, I agree. TOS-R was correct to revise such gaffes. The NITPICKER'S GUIDE FOR TOS, TNG and DS9 books did a good job of exposing such production shortcomings. The written words of the script are always interesting to explore.