I think I worded it wrong.
My html files css had:
h1, h2, h3, h4 { page-break-before: always; }
From that html I can create an lrf and an epub with correct page-breaks.
But, if I run epub2lrf on that epub, which itself has correct page-breaks, the resulting lrf will have incorrect page-breaks (only breaks when it finds the "chapter detection words" in the headers).
I'll attach example files.
The lrf was generated from the epub. The epub breaks at each chapter, but the lrf only breaks on chapter headers that contain "book".
Cather, Willa - One Of Ours.epub
Cather, Willa - One Of Ours.lrf