Part combiner: identical combinations of rests and skips

[Dan Eble]

After this commit

  commit 386c1fb71fef99d9ac865ccd1449412c26cef5cc
  Author: Dan Eble <address@hidden>
  Date:   Sun Nov 23 23:55:03 2014 -0500

The rests in this example are not merged when \global is present in
the individual voices.  If \global is removed, the rests are merged.
It is tempting to say that keeping the voices apart when there are
concurrent skips and rests is intended; but then why are the rests
merged even though there are still skips present inside \relative?

For the full example that led to the discovery of this bug, see
http://www.mutopiaproject.org/ftp/MozartWA/KV250/kv250/ movement 1,
measures 22-23.

Maybe the part combiner should be changed to ignore skips again (but
just skips) so that 2.20 will not break scores like this; or maybe it
should be complicated even more so that it properly sees that the two
parts are identical in the rests and skips they contain.

\version "2.19.0"

global = { s1 s1*3 } % tempo, rehearsal marks, bar lines, etc.

fagotti = \relative c' { d2 d | R1*2 | gis2 b }
fagottoI = << \global \fagotti \relative c { s1 s1*3 } >>
fagottoII = << \global \fagotti \relative c { s1 s1*3 } >>

\new Staff {
     \partcombine \fagottoI \fagottoII
}