The heart of the matter: any lens capable of reproducing these gratings has to resolve clearly at least 100 lp/mm, and over a 4x5" field at that. Such lenses exist, but they're not cheap.

There is a huge support and service industry supplying chip-makers with photolithography tools and chemicals. One part of this is makers and suppliers of the 'masks' used to define the patterns which are then projected onto a photosensitive resist with the hugely expensive lenses I just mentioned. These masks are usually chrome on glass (or quartz) and are tough enough to be used for contact printing as well as projection printing.

One standard size is 5" square, which will have a large enough patterned area to contact print onto 4x5 film. Masks are specified according to the size and precision of the line features, and you pay for every step in quality, but your needs are at the low end of what is regular production, and you should be able to get masks made for a reasonable price. I used to typically pay €100 for this size and quality for one-off patterns, but I did the pattern design myself. You may be able to get a local university department to help you out (mask-making is also sometimes a routine service at high-tech incubator parks for startups) and pull the cost down further, or just help with design and ordering so that you avoid re-makes.

Such a mask would work well as a master, which could be contact printed onto lith film to make working gratings for print production.

I calculate that Wood's rulings produce coloured light at around 72-73 off axis. The viewer in the article you posted would either need to include a hefty wedge in the lens in front of the image, or have the peephole way off to the side, or (more likely) illuminate the image with collimated light coming in from the back at 72-73 off axis. The sun will work well if placed at the right angle, but a projected beam might be more reliable, especially in Rochester winters.