[PATCH] Refactoring for mode-dependent addresses

[Paolo Bonzini]

> This patch decreases the likelihood of similar slip-ups in the
> implementation of GO_IF_MODE_DEPENDENT_ADDRESS and fixes this Thumb bug
> as a consequence.  I appreciate this affects a lot of targets but this
> seems like a good clean-up to make.

Can you please commit it to the addressing-modes branch too?  It was on 
my to-do list as well!

Paolo