I run into this problem so often that I've been experimenting with rendering all the PDF pages to png images first, and a program that loads all of them on screen at once.

The upside is that it is extremely fast, there is no loading time because all the images are loaded already.

The downside is that it uses an obscene amount of memory. A 500 page manual could run you 8-10Gb of RAM. Also there is no more text highlighting or control-F.

But the fact that even this "works" as well as it does indicates how much worse PDF readers are at this, especially when you get those scanned documents that still have the full images anyways, with lousy or no OCR. They take forever (multiple seconds per page) to load when your in a PDF in chrome and jump somewhere else in the document. Why not go back to image files for those ones to begin with?