inventory which characters are used from which blocks (this should be easily scriptable: just run through the content files and build a list of used codepoints, then dump that to a file)
determine which free font(s) can act as basis for creating a dedicated subset font
the book itself already uses HAN NOM A and B, so we might be able to just subset out of those.
This would require: