The new Tamron 24-70 with image stabilisation is by all accounts as good as the Nikon 24-70 (with the added benefit of the IS) and should be way better than the 24-120.
If it's true that the Tamron 24-70 with image stabilisation is as good as the slightly heavier and more expensive Nikkor 24-70, then the choice would appear to be a no-brainer.The walk-around purpose of the lens implies that a tripod will not be used most of the time, therefore image stabilisation is of tremendous benefit.
However, I doubt that the Tamron would be as sharp as the Nikkor, but it might be in the centre. I've seen reports that the Tamron has poor edge performance on full-frame.
If a lens is for general walk-about purposes, there has to be some compromise between maximum performance under ideal conditions, and a degree of flexibility that allows one to get an acceptably sharp shot, as opposed to not getting an acceptably sharp shot and perhaps not getting the shot at all.
Certainly in that range of 24m to 70mm, the 24-70/2.8 should produce marginally, but noticeably, sharper resuslts than the 24-120/F4 does within the same range. However, between about 80mm and 120mm the situation is reversed when one compares the uncropped image from the 24-120/F4 with a 70mm shot from the 24-70/2.8 cropped to the same FoV as the 24-120 used in the range of, say, 80 to120mm.
Out of curiosity, I checked the results at Photozone again for these two lenses.
It seems the situation is far more than merely reversed. Referring to the MTF 50 results at Photozone for these two lenses, tested on the D3X, we find that the 24-70 at 70mm is sharpest at F5.6, producing 3706 LW/PH.
At 120mm the 24-120 is also sharpest at F5.6, producing 3465 LW/PH, in the centre. The difference is 241 LW/PH, or a 7% increase in resolution for the 24-70, at 70mm.
However, if we compare that increased resolution with the increased magnification of the 120mm lens compared with a 70mm lens, we find that the 120mm lens represents a 71% increase in magnification both horizontally and vertically. In other words, if both lenses were the same quality, the 120mm lens would deliver 71% more resolution than the 70mm lens, comparing equal FoV shots taken from the same position.
However, the lenses are not the same quality. The 70mm lens delivers 7% more resolution per unit area of sensor (on the D3X). Subtracting that 7% advantage we get a nett resolution advantage of a whopping 64% for the 120mm lens (71%-7%).
It seems clear that in circumstances when you need a 120mm lens for the desired composition, but your maximum focal length is only a 70mm lens because your walk-around lens is the Nikkor (or Tamron) 24-70/F2.8, then the 24-120/F4 would deliver 64% more resolution at 120mm, which is a far greater increase in resolution than the 24-70/2.8 provides within the range common to both lenses.
At 24mm the 24-70 has a resolution of 3988 LW/PH at its sharpest aperture of F4. The 24-120 at 24mm and F4, also its sharpest aperture, has a resolution of only 3784 LW/PH, 204 line pairs less.
What's the significance of this? An increased resolution of 204 line pairs over 3784, represents an increase in resolution of a mere 5.4%, which pales into insignificance compared with the 64% increase in resolution of the 120mm lens compared with the 70mm lens of the 24-70/2.8, when that image is cropped to the same FoV as the 120mm lens.
I suppose one could argue that the wider aperture of the 24-70 provides a shallow DoF advantage. On the other hand, the VR of the 24-120/F4 will often provide a qualitative advantage for hand-held shots because it enables the use of a slower shutter speed which in turn enables the use of a lower ISO with less noise.