No, it's not a container like a Quicktime or MKV. But it can contain image data that can easily be parsed out by programs that don't necessarily render other parts of a PDF file. Both bitmap/raster and vector based.

Definitely not ideal for an intermediary - at least for raster based images. It works fine as an intermediary to store vector images and text however, though not the ideal solution for text IMO.

I generally keep source files around for everything, but I will often publish (collections of) both raster and vector-based images to PDF, as well of course text-based data as a final (submission/presentation/distribution) format.

But once Doug is done his edits, storing high-quality images combined in a PDF is an absolutely fine solution for moving all those images around in a single file. It's a much more portable solution to holding the TIFF or JPG data.

Though if the content is truly all text based, the ultimate solution is to run OCR against it to produce a clean text document which can be stored much more efficiently. A nice little program on the Mac, PDFPen includes OCR built in, so when you load up an image-based PDF, it can run OCR on it automatically if you so choose. There are obviously enough solutions available for Windows to make this practical there too, even if run first against the TIFF images.