Your explanation as to how non-Nikon pano software stitches images together was very helpful. At least now I understand why such software cannot "re-assemble" the pano in the original NEF format. As to shooting the pano to create a "faux" HDR effect, I am actually maximizing the large amount of data captured in RAW to then convert the same image at 2 (or more) different exposure levels before then blending them back together. With a simple brush I can then "expose" only those parts of the sandwiched image for which the RAW image was processed to the correct exposure (say sky vs. foreground, for example). I've tried this blending with TIFF format files on a large pano I created, but only with limited success. I believe that the headroom inherent to RAW data results in a better, ie. more dynamic color range in the blended image - at least the TIFF composite looks flatter to me on my monitor.
What you suggest is what I do for singular images. But this blending only works well when the multiple images align perfectly - just like the "layers" in PS. The automated stitching process in CS3 does not necessarily create an absolute alignment match over multiple panels for successive images processed independently. In the absence of a stitching software that leaves the RAW format intact, I've tried the process you've suggest without much success.
Thanks to both of you for your replies!