Revert change 1509 that flush ICs when adding setters on an object or...

Revert change 1509 that flush ICs when adding setters on an object or
when setting a __proto__ to an object that holds a setter.

This seems to cause a major page load regression, so we need to tune
the clearing.

Committed: http://code.google.com/p/v8/source/detail?r=1548

[Christian Plesner Hansen, 2009-03-20 20:08:59]

Lgtm.

We knew this might be the case and there is a number of ways this change can be
improved.  I'd like to do some experiments, any way for me to run whatever
showed the regression on a local machine, or something that would show the same
regression?