If this was a well paying job, I would do things quite differently, especially at the start.
1. I wouldn't take a photo of it with a Canon G9 and save the image as a fairly highly compressed, 8 bits per channel JPG. Rather, I would scan it on a good scanner at 16 bpc, and save the result as a lossless compressed TIF. The reason is fairly obvious if you peek ahead at the masks I developed -- look at all the horrible JPG compression artifacts.
2. My next step would be to bring the TIF into ACR and correct the image for lens (ie, pillow) and geometric (eg, keystone-like) distortions. Doing this instead of immediately masking or cropping will guarantee that you aren't throwing away any areas near the edges of the image. If you had scanned this instead of re-photographed it, this step probably wouldn't even be necessary. BTW, notice that my corrections were designed to restore the rectangular border of the print to a rectangular shape, NOT to correct the keystone and other distortions present on the original image. That can be done later, if it is deemed appropriate from a conservation POV.
3. At this point, I would make two copies. One with only the distortion corrections. This will be be used later. The second copy has the distortion corrections, but also has contrast, clarity and vibrance / saturation cranked all the way up so that it's easy to automatically distinguish between the different types of damage to the print.
The reason for doing so is that the areas that faded to (a) a cool tone have a very different chemistry, and hence, a very different noise / grain character, as well as very different contrast compared to areas that (b) changed to a dark sepia, (c) areas that changed to a light sepia, (d) areas where the surface of the area was pulled off and now stand out as yellow-orange. For the best combination of noise reduction and recovery of detail in each of these areas, they must be treated separately.
Here is the 2nd version, followed by the various masks I created:
1. Distortions corrected, contrast, etc all cranked up to distinguish the various types of damage.
2. Mask for the cyan-blue areas:
3. Mask for the yellow-orange areas (ie, where spots flaked or peeled off the print):
4. Mask for the dark sepia areas:
5. Mask for the light sepia areas:
More later about how one uses these various masks to extract maximum detail (at minimum noise) from the image.
Tom