Hi Bob. This is a touchy subject. As you know, I am very much a code minimum type of guy when it comes to enforcement. I don't use the crystal ball and say "what if this happens", and I am a far cry from a "good ole' boy".
One of the questions you ask is if 250.4(A)(5) is applied, why is 250.122(B) in the code? That is a great point. I think 250.122(B) is quite general, meaning that if you have so much impedance on your circuit that you have to upsize the ungrounded conductors, you will also have much impedance on the EGC, so it also needs to be upsized.
Now lets consider an application where in my opinion 250.4(A)(5) might trump the rest of article 250. Lets say I have a motor that is protected by an inverse time circuit breaker. Lets say such motor draws 24.2 amps (3 phase 208V, 7 1/2 HP, table 430.250), and my inverse time circuit breaker is therefore 70 amps (24.2*250%, next size up=70). I can use 12 AWG ungrounded and 12 AWG EGC for this circuit, because 250.122(A) tells me that my EGC need not be larger than my ungrounded conductor. Now, if the instaneous trip of this breaker is, say, 300 amps, and the fault current produced by this circuit doesn't create that, I have violated 250.4(A)(5). This would require me to increase either the size of the ungrounded conductors, or the size of the EGC, or both. The note to table 250.122 would override 250.122(A)'s requirement of not exceeding the ungrounded conductor size, because 250.122(A) points us to that table, and that table contains a note that IS part of the table.
While I agree this is probably a rare occurance, it is my opinion that 250.4(A)(5) is referenced by the table for just such as event, regardless of how improbable it may be.
When you were asking how I would base my decision, I would have to base it in the instananeous trip level of the OCPD, which will vary by the installation, of course.
I also think that the intended result of the performance based requirement is in fact spelled out. From the 2005:
(A)(5) Effective Ground-Fault Current Path. Electrical equipment and wiring and other electrically conductive material likely to become energized shall be installed in a manner that creates a permanent, low-impedance circuit facilitating the operation of the overcurrent device or ground detector for high impedance grounded systems. It shall be capable of safely carrying the maximum ground-fault current likely to be imposed on it from any point on the wiring system where a ground fault may occur to the electrical supply source. The earth shall not be considered as an effective ground fault current path.
If the conductor has such impedance that it will not initiate the OCPD, then the first bolded item is not satisfied. If the conductor has such impedance that its short time withstand rating is arrived at before the OCPD is inititated, then the second bolded item is not satisfied.