I think the document is scanned from a paper copy. The pixels shown on the screen is from the scanned image, but the text for copy-and-paste was created by OCR

I think the document is scanned from a paper copy. The pixels shown on the screen is from the scanned image, but the text for copy-and-paste was created by OCR when the document was scanned and prepared. You often see this with pdf files, I guess the software used to create them has OCR functionality built in. If you use "inspect element" in your web browser you can see how it is displayed. Each page is a <div> which contains first a big image (with the scanned document page), and then a <span> for each word, which are positioned using absolute positioning to be placed over the corresponding word in the image. The <span>s each have "color: transparent", so you can't see them but you can still select them.