(another) bounding box bug in RC1

[Cyrille Chepelov <cyrille chepelov org>]

Hello all, 

  no time at all this evening; however, I've noticed there's a bug in
samples/grafcet.dia (!); the bounding box of two transitions (object number
78) seems incorrect. When emacs'ing them to saner values, it seems the
problem re-appears as soon as the diagram is reloaded. Probably an uninit'ed
variable somewhere in the transition's bounding box computation.

  I'll investigate (valgrind, printk and all), but in the mean-time I post t
his for the record.

	-- Cyrille

-- 
Grumpf.