Braille chapters

[Rob Beezer]

For Alexei K. Your routine for busting up a braille book into separate files
for each chapter just got integrated into the Python routines. Now Al M.
suggests we don't say "Chapter" in the heading. Making landmarks a bit harder
to find.

We know in the Python if the publisher has requested chunk level 0 or 1.
Suppose we write out "CHAP" from the XSL in all cases, a marker that will never
occur in a BRF. Chunk level 0, we replace it by two spaces (preserving
centering from liblouis). For chunk level 1 we use this in the regex to locate
the chapter start, and then replace it after it has served its purpose.

We get identical BRF in each case. We do not mess with line-breaks etc. We do
force line-breaks more often than needed, but it will be consistent. I guess we
could use "CH" as the signal. Or maybe just "C" and depending on parity we
"fix" an unbalanced centering with no apparent compromises.

Workable?

Could be sustainable if we need a scheme to bust into sections...

Rob