* lily/parser.yy (simple_markup): replace THIS with PARSER.

* lily/parser.yy (simple_markup): replace THIS with PARSER.

* ly/Welcome-to-LilyPond-MacOS.ly: include in LilyPond, so version
number stays up to date.

* buildscripts/output-distance.py
(ComparisonData.create_html_result_page): new routine: summarise
results in HTML page with images.
(ComparisonData.create_text_result_page): create summary text files too.

* GNUmakefile.in (web-ext): package .signature files too. They
compress well.

* scm/stencil.scm (write-system-signature): typo.

* GNUmakefile.in: create web-root/VERSION to help upload scripts.

* scm/stencil.scm (write-system-signature): better escaping.
only write signature for music systems, not for markup systems.

* scm/framework-ps.scm (output-framework): dump-signatures support
for plain output.

* make/lilypond-vars.make (LILYPOND_BOOK_FLAGS): add -ddump-signatures.

* scm/framework-ps.scm (write-preamble): redefine mark_URI to nop
for -dno-point-and-click

* lily/grob.cc (get_print_stencil): don't read
point_and_click_global to determine whether to set grob-cause.

* buildscripts/output-distance.py (main): new function.
(ComparisonData): new class.

* lily/system.cc (get_paper_system): remove 3 layer limit.

* buildscripts/output-distance.py (SystemLink.distance): new
file. Compare signatures.

* python/safeeval.py (SafeEval.visitUnarySub): new file.

* lily/grob-array.cc (remove_duplicates): rename from
uniq(). Sort before calling uniq() so it actually works.

* lily/system.cc (get_paper_systems): uniq all_elements_ , to
prevent duplicated broken items in output.

* scm/framework-eps.scm (output-classic-framework): only dump if
dump-signatures is set.

* scm/lily.scm (define-scheme-options): add dump-signatures option.

* scm/stencil.scm (write-system-signature): new routine: write
python parseable signature of a separate paper system.

* lily/stanza-number-engraver.cc (process_music): use is_markup()
iso is_string().
include header.

* stepmake/aclocal.m4: STEPMAKE_PYTHON(): clear cached value since
arg 2 might point us to a new binary (patch by Karl Hammar)