Spacing staves with dynamics between; issue 1668

Spacing staves with loose lines between.

Distinguish the minimum distances between 'spaceable' staves
from those between lines such as Lyrics.

[Keith, 2011-05-27 17:37:06]

Consider each minimum-distance and padding constraint, and choose spacing that
meets the most demanding constraint.

Previously some distances were added together by mistake in the within-system
spacing; disagreeing with the documentation and with behavior for system-system
spacing.  Thus the Lyrics minimum-distance is bumped up to 3.0, to match
previous behavior when the old minimum-distance=2.8 was always augmented by
padding=0.2.

Other small changes in regression tests are due to getting exactly the spacing
the input files asked for.

[Keith, 2011-05-27 18:02:23]

Joe just pushed half of this, fixing 1442, so never mind for now.  
I'll merge and repost next weekend.

[joeneeman, 2011-05-27 19:42:47]

On 2011/05/27 18:02:23, Keith wrote:
> Joe just pushed half of this, fixing 1442, so never mind for now.  
> I'll merge and repost next weekend.

Sorry, I didn't realize we were competing for this issue. But it seems like
there's still an outstanding issue in the handling of minimum-distance together
with padding?

[Keith, 2011-05-27 19:47:15]

On 2011/05/27 19:42:47, joeneeman wrote:
> Sorry, I didn't realize we were competing for this issue. But it seems like
> there's still an outstanding issue in the handling of minimum-distance
together
> with padding?

I just put the regtest for this patch as issue 1668.  I flagged that issue as
"Started" and linked to this Rietveld, but feel free to adopt it (and put
yourself as owner) if you are thinking about spacing.

[joeneeman, 2011-05-28 07:00:55]

lgtm

http://codereview.appspot.com/4536088/diff/15003/lily/page-layout-problem.cc
File lily/page-layout-problem.cc (left):

http://codereview.appspot.com/4536088/diff/15003/lily/page-layout-problem.cc#...
lily/page-layout-problem.cc:543: // using a scheme similar to the one in
set_column_rods.
I'd prefer to leave this comment in

[Keith, 2011-05-28 18:42:02]

On 2011/05/28 07:00:55, joeneeman wrote:
> I'd prefer to leave this comment in
Oops; comments from your last patch are back in.

Re-reading our comments, it is clear that we always want to include
minimum-distances between adjacent lines when computing
get_minimum_translations.  Adjusted patch is up.

The inaccuracy described in your comment happens with lines M v and U where the
v can rattle around between M and U.  We fail to distinguish the distance
between rattling v and U from the actual minimum v-U distance.  Omitting M-v
minimum-distance just gives more room to rattle.

To fix 1442, we needed to ignore only the minimum-distance between
*non*-adjacent already-placed Staffs M and U while distributing loose-line v.