the htmltopdf.js library renders all PDF text as images instead of
vectors, and Go does not have a reliable native library for generating
PDFs that is actively maintained. The best solution appears to be to
make a system call to wkhtmltopdf, but now there is a hidden package
dependency that is not checked at build time and exists in another
process. Maybe in the future a C library can be used and linked with the
Go binary.