I am looking at the excellent book EPUB 3 BEST PRACTICES by Matt Garrish and Markus Grylling in the section "Page Numbering."
First, we need to describe what the function of the
<nav epub:type="page-list" ..> element is supposed to do. Some reading systems (specifically Readium) have the ability to search specifically by page number (presumably printed page number). IN Garrish's book this feature is called "page jump functionality."
Only if a reading system provides this functionality, then it would be helpful to include an ordered list within the nav "page-list" type. So nav epub.page-list is an optional method for providing an alternate navigation/search method.
Important: the only constraint on page-list type is that it must contain an ordered list. For this reason, the semantics for LI are interchangeable; there's no special requirements or constraints on the value. If the writers of the standard wanted to constrain the values of the ordered list items, they would have created an additional unique element with the nav page-list element.
So yes, these references are arbitrary -- you could use letters of the alphabet, Roman numerals; it does not matter.
Later the book mentions providing anchors for page numbering.
Presumably for the example you used, you would switch id=page752 on the georgia.xhtml page.
(The book also describes the pros and cons of using css to show/hide these page indicators throughout the text).
A few things worth noting: First, I don't even know which reading systems support this feature of search by page number. Second, I am skeptical about whether this is implemented on Kindles (I could be wrong...) Third, I suspect the ideal way to implement references to printed pages is by letting a tool like Adobe Indesign auto-generate these things.