As far as I know, fixed layout
epub3 is being used mainly for children books, manga comics and educational textbooks with lots of pictures etc. For documents usually produced with LaTeX, ie. text based documents with floating tables, graphs and images, reflowable epub is better, as the reading device may decide itself where to break pages, depending on the size of the device. For some articles about issues with fixed layouts, see this and this articles.
For the question about ease of fixed layout epub3 creation from LaTeX, no it would be not easy. Lots of
.html convertors exist, which can be used as base for
epub conversion. In fact, there is tex4ebook, tool which use
tex4ht convertor to translate LaTeX directly
mobi formats (note: I am the author). for fixed layout, special configurations would have to be provided and some commands would need to be inserted in the LaTeX source.
In my opinion, reasons why not so many ebooks produced by LaTeX exist are that:
There is really bad support for math in ebook readers. epub3 readers should support
mathml, but in reality, they don't. In the best case, they use mathjax internally for math rendering, but it was painfully slow when I tried one of such apps on my android phone. this may got better in the future, I hope. You may test some
epubs from Frédéric Wang or Glenn Barnich to see how well your reader support
mobi, math must be included as images, which looks really horrible especially for inline math - characters aren't on the baseline, a different font than in the document is used and if the images have a different resolution than the device, they may be too big or deformed (see this for some experimental solution).
ebooks look much worse than
pdf. While they have some positive features over
pdf, like the possibility to resize text or change the font, ebook readers use html viewers for rendering and the typography they produce really doesn't satisfy standards for someone who uses LaTeX to produce typographically perfect documents.
.html convertors are hard to use and they have sparse documentation, so many people stop trying when they encounter some problem (this is probably the main reason).