[PEAK] A couple more Component initialization cases

[Phillip J. Eby]

At 12:17 PM 8/14/2007 -0400, Phillip J. Eby wrote:
>No, it's still me that's wrong here.  The new initialization rules 
>are still borked.  The new rule works fine for initializing 
>circularly-dependent rules, but not for overriding rule defaults.

This should be fixed in SVN now.  The fix is a really ugly hack, but 
it's not worth making it neater as the new algorithm in 0.6 should 
make this irrelevant.  (Because written values will always override 
calculated values.)