Actually, you could try any example.
lilypond-bin input/regression/accidental-double.ly
sort accidental-double.tex > sortedlines
sort -u accidental-double.tex > uniquesortedlines
diff sortedlines uniquesortedlines
29d28
< \def\lilypondpaperunit{mm}%
49d47
< \lyitem{ -4.0000 }{ 12.5729 }{\magfontSDEXBomMMBo\char47 }%
54d51
< \lyitem{ -6.0000 }{ 14.9229 }{\magfontSDEXBomMMBo\char150 }%
61d57
< \lyitem{ -7.0000 }{ 8.5729 }{\magfontSDEXBomMMBo\char144 }%
63,64d58
< \lyitem{ -7.5000 }{ 21.8437 }{\magfontSDEXBomMMBo\char47 }%
< \lyitem{ -7.5000 }{ 23.1437 }{\magfontSDEXBomMMBo\char62 }%
71d64
< }%
As you see, the clef, the time signature and the key signature are
typeset twice
(the two last duplications belong to the duplicated note of this
particular example file).
I have no idea why it happens.