[developers] PET regression fix

[Bart Cramer]

Dear developers,

I just discovered a bug, introduced when the -cp option was introduced
in the cm branch (rev 722), and is also present in the current main
branch. The bug is present due to an incorrect default for -cp. The
bug was not discovered before, as the standard setting (400 passive
edges per chart cell) is large enough to only affect sentences with
large ambiguity.

A patch for the 'main' branch has just been committed (rev 766). If
necessary, I can also do it for the 'legacy-cm' branch, as the bug is
present there as well.

My apologies for any inconvenience.

Best,
Bart.