We are actually in agreement, I think. I concur that in some cases line database attribution might be a prerequisite for jurisdiction, as I've written here http://ssrn.com/abstract=1139174. Take Loizidou as an example: if jurisdiction means effective overall control of an area (northern Cyprus), and the question is whether Turkey had such control, then it must be established that persons whose acts are attributable to Turkey exercised such control. In Loizidou this was not in dispute, as it was the regular Turkish armed forces who were in control of northern Cyprus. The same is with Behrami, as you mentioned - it was perfectly proper for the Court to see whether the acts of French troops in Kosovo were still attributable to France, before it could say whether France had control over a part of Kosovo. It's just that it got the attribution question so spectacularly wrong.
Now, why is it that I think that the Court of Appeal was not ruling on attribution: because this argument was actually made before the High Court, which clearly rejected it (see para. 29 CA, para. 79 HC). The European Court case that was on point was Drozd and Janousek v. France and Spain, where the ECtHR found that the acts of French and Spanish judges that served as judges in Andorra were not attributable to France and Spain, as they were put at the disposal of the state of Andorra. This is of course Art. 6 ILC ASR attribution. You will see that the Court of Appeal makes no mention of Drozd, other than in para. 29. They are therefore not making the Drozd attribution argument, they are ruling on the basis that Art. 1 jurisdiction requires the exercise of legal authority, which the UK did not have independently of Iraq.