sort.  Instead of caching the output of QString::lower() go a level
deeper in the string comparison logic and cache the value of
QString::local8Bit().  I think this is about as fast as it's going to get
for the moment, but it's now down to 4.8 seconds (on my machine) for 12,000
items -- compared to 94 s when I started profiling with this data set.

svn path=/trunk/kdemultimedia/juk/; revision=242706