MusicXML regression tests

[Phil Holmes]

A guy who uses the same notation software as I do has been working on a 
MusicXML importer and wanted some sample test files, so I pointed him at our 
regtests.  He commented the following:

I just ran all the lilypond musicxml regression files through my
program, just watching for any musicxml errors.  I found the
following, which you are welcome to let lilypond know about if
you like, or ignore if you don't!

21f - in a "pitch", has an "alter" after an "octave" when proper
      musicxml is to have it before

33a - has a "grouping" under "direction-type" when proper
       musicxml is to have it directly under "measure"

71f - line 246 is an extraneous "lyric" under "harmony"

74a - had an empty "figured-bass" (on purpose for some reason)

I've made appropriate changes to my versions of the regression
files, and have no plans of changing my program to accommodate
invalid Music XML!  More details to come soon...

I'm not expert enough to know whether these are bugs.  Can anyone comment?

--
Phil Holmes
Bug Squad