FrameMaker is IMHO the best example, though there are others like Interleaf and 3B2.
Ding ding ding, we have a winner!
Interesting fact: The original web client, Nexus, was not a web "browser", but an integrated GUI HTML browser/editor, as were the LibWWW API, and all of W³C's subsequent testbed clients. The intended paradigm was using a GUI to read and write webpages (and yes, style sheets did exist, though they were very limited in terms of flexibility), while HTML was never normally intended to be seen by human eyes nor written by human hands.
The idea of a read-only "browser" was the accidental result of producing the supposedly specialized Line Mode Browser, the first CLI client, and also the first crossplatform (Nexus was NeXTSTEP-only) client. Since Line Mode Browser was how nearly everyone outside the NeXT bubble first experienced the web, combined with the also crossplatform CERN httpd server, this meant nearly everyone who wrote clients like Viola and Mosaic thought the "normal" way to use the web was to write HTML by hand in a text editor, and view it in a read-only browser.
CERN/W³C, in retrospect, considered this a critical failure on their part.
Something like an imageboard honestly shouldn't function inside a website, with a better fit for the web format more closely resembling a wiki (e.g.: heavy reliance on powerful automated transclusion, Boolean operations, etc.) to reuse, generate, and organize content based on machine-readable semantic tags.
For something more like an imageboard, a database-centric client API like USENET would probably be better than a website to begin with, though on top of that transport/storage layer, a markup-based semantic and presentational standard would be superior to plaintext.