TODO:
====

*) tidy up getopt issues - distribute GNU getopt with wav2cdr and make it
   default; give option for mygetopt

*) For
      wav2cdr -r Krebspolka.wav -w Krebspolka.cdr 
   the output filename should be Krebspolka.cdr, not Krebspolka.cdr.%02d
   (i.e. when no cutting is active)
   To get a filename with %0d, use "--cut 0 0"
   However, to get a filename without %02d now, use output redirection.

*) When data is read from file and cutting is active, seek operations should be
used for cutting because, depending on how much is cut out, it will be
considerably faster.

*) Output total bytes read.

*) Operate on more than 1 file, if given on command line. This does not work
with all operations (--cut) but does with others (--silenceinfo). All
initialised static variables need to be abolished - the result has to look
pretty similar to a shared library.
This could afterall make sense with --cut anyway. Resulting filenames to be
decided upon - overwrite existing file (by creating new, then renaming)?


Ideas:
=====

Store raw data in files which have an extension denoting the endianness. For
example, extensions .rawbig and .rawlittle (or .rwb, .rwl on less capable
systems). The extension can then be used as information on the byte order in
the file when reading it again - this would make errors less likely.

When cutting, specify a cut to be discarded. This would avoid storing silent
intervals between actual tracks.
What syntax to use for this??

Display seconds in --silenceinfo rounded instead of truncated.
MANUAL: ARE SECONDS NOW DISPLAYED ROUNDED ELSEWHERE???