Some quick thoughts.
First, regarding continual updates… One option is to use the wiki’s Time Machine to archive the pages prior to the book you’re reading. You’ll get a solid database of information and (mostly) avoid spoilers. You’d still face a small problem with outdated wiki pages, however.
As for the scale of the task… You’re looking at somewhere between 5,227 and 21,070 pages to download. (I’m using this as source: https://coppermind.net/wiki/Special:Statistics) The lower bound is if you download solely content pages, while the upper bound includes stuff like redirects, disambiguation pages, talk pages, user pages, etc. This is sizable, but still quite manageable. Way better than trying to turn Wikipedia into an EPUB. However, I’m not sure how well your Kindle will do with searching such a large “book” for text.
I’d recommend doing some tests on a smaller scale first. Download 100 pages (preferably larger, wordier ones), see about merging them all into a single book, and see what works and what doesn’t. Can you effectively search for the text, or do other pages drown it out? (E.g. If you’re searching for Kaladin’s page, are you going to have to wade through a hundred mentions of him on Syl’s page first?) Will you need some kind of table of contents? If so, can that table of contents page scale upwards once you have thousands of pages to deal with? Do tables work or do they break on your Kindle? What happens if you tap on a redirect? There are lots of little questions to address!
This tool, https://github.com/hauke96/wiki2book?tab=readme-ov-file, handles the conversion process for multiple Wikipedia pages into a single EPUB file. I haven’t given it a try, but you might want to take a look at it and see how they’ve managed it. It might give you some good hints and ease the process along.
Best of luck!