I've reproduced the problem to some extent and I think I've determined the cause and a solution.
First off, I don't have a 10 stop ND filter - sorry guys, I'm not dropping 100$ just to do this test and then watch the filter collect dust on the shelf since I have absolutely zero use for such a thing, but I do have a lens cap....
I have always thought the problem might be a light leak, so my test setup was as follows:
D300 on tripod/ballhead, set in landscape position
24-70/2.8 lens set at F/9 and 50mm
long NR set to "on"
Shutter set to bulb
remote release timed exposure of 4 minutes
Instead of a 10 stop ND, I put the lens cap on the lens and put the lens hood on the lens as I normally would.
Took a 250 w/s Dynalite strobe I had lying around, mounted it so the unit aimed down at approximately 40 degrees straight on into the lens/camera.
Center of flash tube was 3 feet from focal plane of lens/body.
Set the monolite at 250 w/s and turned the modeling light on full.
Ambient light reading at that distance for that ISO was 1/250 at F/16, which emulates strong daylight.
In addition, during the 4 minute exposure I fired the strobe at full power (250 w/s) 30 (yes, thirty) times, which I do believe is considerably brighter than any daylight exposure you're ever likely to encounter on this planet.
My first frame showed a narrow horizontal area along the top left edge of light leak/flare - from the left edge to maybe 1/4 to 1/3 inwards. Not very tall, perhaps 3/8 of an inch on the back LCD screen. The rest of the frame was pitch black.
I then redid the test with just the ambient light from the modeling light (did not hit the strobe 30 times at full power) and got a weaker version of the above.
I then covered the lens barrel with a dark washcloth and repeated the first test (4 minutes ambient, 30 hits with the strobe): result: pure black frame.
Then I removed the dark washcloth and covered the plastic cover over the distance display on the top of the lens barrel with a double folded piece of gaffers tape and redid the same test (4 minutes ambient, 30 hits with the strobe). Result: no flare, pure black frame.
Obviously while I can't exactly reproduce the ND filter scenario, I'm deducing that there is a light leak around that plastic distance display in these conditions that's causing the problem.
While nothing I say will make the 24-70 haters ever like this lens, I would say if you want to do this sort of long exposure/daylight shooting, I'm betting a double layer of black masking tape over the plastic distance scale would solve the problem.
Hope this helps....
(and perhaps someone can also verify my test to see if they come up with the same thing)
ps: I apologize for not providing samples: I don't have a hosting site at the moment.