No, I don't see it that way. From the opening post, It seems to me that the bonding jumper in question is not being routed "with" the conduits, but merely "to" the conduits. The fault path that runs "with" the conduits is the conduit itself. The largest bonding jumper ever required by table 250.66 is 3/0. In my opinion, if individual bonding jumpers were to be run to each conduit, and if the equivalent conductor size in each conduit were 1100 kcmil or less, then there might be an argument for using the section of code that iwire quoted to downsize those jumpers to something less than 3/0.
Notice the significant differance in sizing rules for parallel conduits on the line side of the disconnect as opposed to section 250.122(F)(1) for conductors on the load side of the disconnect.
