Fix base attachments for slurs at linebreaks

Fix base attachments for slurs at linebreaks

- Acknowledge BarLines, but do not acknowledge SpanBars, since they
contain references to BarLines in all staves.

- For the base attachments (endpoints) of slurs that occur at
linebreaks, we want to calculate the X-offset based on one of the
slur's encompass objects:  the "rightmost" for beginning of line, and
the "leftmost" for the end of line.

- Add "padding" for the slurs, so they do not touch the edge of the
leftmost or rightmost encompass object.

TODO: add more regression tests.

[hanwenn, 2009-08-23 17:23:09]

http://codereview.appspot.com/109071/diff/39/46
File lily/phrasing-slur-engraver.cc (right):

http://codereview.appspot.com/109071/diff/39/46#newcode147
Line 147: if (!g->internal_has_interface (ly_symbol2scm ("span-bar-interface")))
this will work in some cases, but is specific to the ordering of the prefatory
material and the existence of the barline.  It would better to catch an object
that has references to all of the breakable items.

http://codereview.appspot.com/109071/diff/39/48
File lily/slur-scoring.cc (right):

http://codereview.appspot.com/109071/diff/39/48#newcode411
Line 411: }
fold this code - left and right are symmetric.

http://codereview.appspot.com/109071/diff/39/48#newcode577
Line 577: x = (d == LEFT) ? (x + 0.25) : (x - 0.25);
dont use conditional

  x += -d * 0.25;

http://codereview.appspot.com/109071/diff/39/49
File lily/slur.cc (right):

http://codereview.appspot.com/109071/diff/39/49#newcode197
Line 197: - Is there a better way to accomplish this?
I think you should first make a test case that contains all combinations of key,
slur, clef, barline etc. at linebreaks, and go from there - hardcoding bar lines
here seems too limited