Bug in 2.17.97, acciaccatura causes blank staff lines.

[Helge Hafting]

> I am not top-posting.

Using an acciaccatura blanks out the staff lines of another Staff.

My minimal example:
\version "2.17.97"
score {   
    \new PianoStaff
    <<
      \new Staff = "up" \relative a' {
        a2 e2
        \mark \default 
        %Remove the mark, and the error goes away!
        
        %Add more notes here, and the problem also goes away;
        %But in a more complicated piece (Grieg's concerto
        %in A minor) the problem does not go away so easily
        
      }
      \new Staff = "down" \relative f' {
        a2 e2
        \acciaccatura e8 a1  
      }
    >>
    \layout{}
}

Compiling this yields:
Warning: Found infinity or nan in output.  Substituting 0.0
warning: Found infinity or nan in output.  Substituting 0.0
warning: Found infinity or nan in output.  Substituting 0.0
warning: Found infinity or nan in output.  Substituting 0.0
warning: Found infinity or nan in output.  Substituting 0.0

The upper piano staff does not get staff lines when this bug strikes. There
is 5 messages, one for each line? The notes are there, but unreadable
without staff lines.

The first bar (a2 e2) is not really needed, but removing it yields a voice
that start with a grace note. That has been problematic in the past, but
this is a different bug. So the first bar shows this - the bug happens even
when the grace note is pushed further out.

The minimal example is extracted from Grieg's concerto in A minor, arranged
for 2 pianos. There, I work around the bug by adding grace skips in all
other voices. Having to do that is tedious enough with the 4 (sometimes 6)
voices you get with 2 pianos. 

The concerto arranged for orchestra is something I don't feel like working
on as long as all-voice grace skips are necessary. Orchestra has too many
voices for this sort of workaround. So I hope this bug will attract some
interest.  2.17 is already much better than 2.13 was, in that 2.13
absolutely needed grace skips everytime. But now, this happens only
occationally. (Or if generating MIDI, but at least these skips are mostly
not needed for the engraving.) Thanks a lot for the fixes already done!