ok the horrible design constraint seems to be that epubs are actually zipped folders containing html files, and those html files are displayed sequentially instead of stitched together into one long text. and i guess adding page-level continuous scrolling would make chapter-level continuous scrolling really janky? meh
on the plus side: epubs are zipped folders containing html files it turns out! i could literally just code up my own epub unsucker in python!
i have no quarrel with hardware epub readers btw, they're just kinda like books and books are fine