I am afraid you all are chasing windmills here. What thecraw has found on his website there, for example, looks like perfectly normal MRC-compression artifacts. I explained how it works
here. The software that created and compressed this PDF simply used an algorithm that isn’t perfect. It tries to seperate the text from the background, and it does not do a perfect job: Sometimes, parts of the text can be found in the background, and in other places, parts of the background will be in the text layer. This is nothing unusual.
For your edification, I took the PDF and re-compressed it using a better piece of software, using a better MRC-compression algorithm. You can find it here. Please open it in the viewer of your choice, and play with the layers (this one really puts the different parts into seperate layers). Note that even though it does a better job at seperating the text from the background (for example, the signatures end up in the text layer, not the background), it still is not perfect.