This set back is common especially with some seismic requirements.  I have seen it from time to time, but it is usually a 1/2" for the stuff I've seen.  I don't know exactly where it comes from, as in what document calls for it.  When I spoke to a detailer they said it was on the contract drawings and thats where they have gotten it.
Now as far as fabrication, it is common to have the fitter or welder miss this note.  It is usually because 99% of the time the other projects don't call for this set back, then everyone get's in the mode of doing the same ol thing they have been doing for months or years, then as soon as you throw something new at them, they either miss it, or freak out and can't believe they have to stop a 1/2" back in my case.
Now when I get drawings that call for it, I highlight it to make it obvious, then more importantly, I ask/ make the fitters measure down and draw a line 1/2" from the end to the hopefully the welder can see this and stop or start at the line and not go past it.
Will it still happen, yeah, but it eliminates or reduces the chances.
When they do exceed it, I call the engineer and ask them if it's ok that we did this on X amount of stiffiners and so far they have always said it is ok as is.  The hard part is you can't really efficiently grind/ remove the end of the weld properly.  Its too tight, you reduce the throat of the weld next to the removed area, personlly in my opion I think its better to leave it as is, than to go grinding it out.  But again you need to get some back up to that when someone comes behind you and calls you on it. This way you can expalin what happened, and what you did to address it.
I would have talked to the detailer like you did, then if time permitted, get with the engineer and expalin the situation.  Some shops find it easier to just address it then and now vs waiting a day or two.  In that case, it gets corrected so your good.
Chris

I just looked at my copy of D1.1-2004 and it shows a sketch of the “hold back” in figure 2.6 and refers the reader to clause 2.8.3.7.

For comparison I looked at a copy of the 1994 edition of D1.1 and a similar requirement is listed for dynamically loaded structures (remember them?). The wording is slightly different for statically loaded structures, but similar so I would have applied the same requirement to both static and dynamic structures.

The concern appears to be limited to welds located in an area subject to tensile stresses. The inspector may not know what areas are subject to tensile stresses especially when a rigid frame (welded moment connections) is involved. Simple framed members (bolted webs only) are fairly simple, the bottom flange is usually the tension flange. If in doubt, the inspector should ask the EOR for clarification. 

What I'm slowly driving at is this is not a new requirement. Fabricators and detailers as well as engineers that keep up with their field of endeavor should know of these nuances. The welder, well we are all too well aware that the welder is our mushrooms of the industry.

How should this be addressed by the TPI, report it and if possible include a photograph of the condition observed, include a reference to the applicable clause in the code and let the fabricator and engineer decide how it needs to be addressed.

The inspector is the eyes and ears of the engineer on the shop floor or in the field. It is not the inspector’s call to make with regards to removing the offending weld or to leave it intact. These decisions should be left to the EOR.

With regards to the "k-area", I was under the impression the welding restriction was for structures meeting seismic requirements only.

Best regards – Al