The officially published ePubs from Sony appear to to be the same way (in the ones I examined).
A couple reasons:
- Table of Contents prefers to point at a file, rather than a location inside a file (goes faster).
- Larger files take longer to load, so splitting it up splits up the 'cost' of loading the book a bit.
- The Adobe SDK has a hard limit of 300KB per HTML file in ePub (uncompressed). Without splitting the book into files, you'd hit this limit with an awful lot of books, since you could hit it with about 160 pages of text.
All this taken into account, it makes sense to split it on chapter boundaries and use that file list to build up the Table of Contents from. If you need to force a page break, it is also a good place to split the text up (up to the judgment of the guy/gal packaging the ePub), since a break between files is treated like a page-break by default.