Some of the paths for things like the textfilt/defn stuff isn't correct yet

The pstogif tools aren't completely installed

The installation stuff in general needs more and careful testing (so as not to
loose the working versions)

The html.def etc files still seem to live in ~/textfilt !

Doctext needs a way to generate index entries.

Mapnames:
       1. Allow case-insensitive match (done), but..
       2. Need to allow replacement text to be original (with its case).
          Currently, it uses the text as it appears in the map file, which
          has no way to say "use original".  That's probably the correct option.

Tohtml changes needed for MPI-3 document:
       1. Handle initial section level that is not nested (e.g., section before 
       chapter)
       2. Handle "*" in section (e.g., no number headings)
       3. obeylines command/environment (behave sort of as verbatim; should 
       this require better handling of active characters) - DONE (?)
       4. texorpdfstring handling
       5. In split mode, it needs to make the html clean and then, on the
       next page, restore any open elements.  For example, things like font
       color and face.
       6. Provide an option to dump index file and to import an HTML file
       to use as the index.  This permits a separate index-processing step.
       7. HTML paragraphs are now <p>....</p>, not a separator.  Need to
       change the generated HTML to match.
       

Bug Fixes needed

Mapnames SEGVs on non-ASCII hyphens (en-dashes?).  These could be in a
comment.

bfort: Does not handle routines with char arguments correctly.  The short fix
is to not generate an incorrect interface in that case; the long fix is
to generate one along the lines of the Fortran stub generator used in MPICH.