I agree with your CERA weaknesses #1 and (the well argued) #2. I would only add this, which you discuss briefly

#3 Uncertainty about depletion rates

CERA on decline rates:
We don't actually look at depletion in terms of an average number across different countries and basins.... we look at it in enough detail that we're trying build up our analysis on a bottom up basis and we're trying, when we can, to get information about specific fields, or, in other cases, where we know about particular reservoirs and different types of reservoirs and particular basins, we make assumptions about decline rates. And so, there's a whole range that we use, from certain Middle Eastern fields where we see depletion rates of lets say 3 or 5%... which can range up to 20% or so for some of the heavy oil fields in Venezuela, where we see depletion rates that, without intervention, are actually pretty high...
Cera on "peak oil":
[what will drive the peak?] ... in the medium term, we don't think these will be subsurface issues. It's mostly going to be subsurface factors that will control the timing and nature of the peak.... [UK past its peak and on an undulating plateau, gradually declining]... [ultimately oil production will peak but we see an undulating plateau out to 2020]... if we consider the behaviour of oil fields and basins and countries and regions to be fractal, we can look at what's happening in particular countries and extrapolate that on a world-wide basis, so we see the undulating plateau existing for a number of decades. And if you look at historical US production, you can see some elements of this shining through.
One final comment. You can not have your cake and eat it too. They seem to make the following 3 incompatible assumptions:
  1. Yearly demand will decline from recent years until 2010
  2. Supply will increase until 2010
  3. Prices will be lower (2007 to 2010) than current levels
Sorry, you can not increase supply, decrease prices and also decrease demand in a world built around economic growth. I'm not an economist but I know an impossibility when I see it.
The "undulating plateau out to 2020" is effectively the same as a peak, and what Colin Campbell was saying a few years ago: The peak has been flattened out by technological/economic factors (lack of drilling rigs on the upslope, "unconventional" oil on the downslope), and we're bumping along the top now. So we won't see a significant decline in supply until 2020.

The main disagreement seems to be that Campbell thinks we're already at the plateaua, whereas Yergin thinks supply can increase for another five years. That's not really much difference, especially considering that Campbell was until recently considered very pessimistic.

Of course, Campbell assumed that Saudi Arabia would peak much later than the rest of the world. If Simmons is right, the decline is already starting.