Re: 13.51 regtests

[Keith OHara]

Phil Holmes <mail <at> philholmes.net> writes:
> 
> Comparing 13.51 to 13.50 with the official regtest checker shows a plethora 
> of minor changes - clearly something in the spacing engine has changed 
> enough to flags lots of errors.  

Horizontal spacing changed for issue 1229.  These all look good to me.  In 
cases 
that you might think regressive, it might help to check for the corresponding 
change from 2.13.24 to 2.13.25.

The reason for the many tiny changes is that the spacing engine looks ahead
for near-collisions in case of moderately-tight spacing.  If any are found,
the spring is slightly lengthened.  So many of the 'softer' springs were 
lengthened slightly, if they lie after an item whose minimum spacing requirement
changed.  (I have a patch to simplify that rule, but am watching for better
examples to of its effect before submitting an item for the tracker.)

Some duplicate tempo marks, we want to have printed, per issue 1205.

On the many performance improvements, I suspect they are thanks to Joe's 
"Optimizations for pure-height approximations".