characters in the user_id_textblock layout, which was backwards,
and (2) add a path reference to "hexdigits" for each new character
layout encountered in the file, so that the text block layouts
are found even if "addpath hexdigits" has not been specified.
There were some other corrections to handling gzipped files that
probably does not apply in practice, where "make uncompress" has
been run prior to the set_user_id.py script.