There seems to be more to it than that.
Adding the ereader2html resulting "book.html" to Calibre and converting it to .pdb, then
converting that .pdb to ePub I get the following internal structure to the ePub file:
And that is it.
Where as, when using Calibre to convert the same "book.html" directly to ePub, I end
up with an ePub file that has a much different structure:
then 18 .png files
Then another 23 .png files
Then another 224 .png files
Then 3 more .png
then 2 .png
It looks to me like maybe if Calibre could have made an images folder and provided
the index html when it did the conversion directly, it could make an ePub ebook that
wouldn't be seen as a "File To Large" to the JBL.