Code 137 is the Code for a 85mm f/1.2L, so it should use the double cross. Apparently, all the Sigma and Tamrons also use this code.
Thanks for the Link. This means that the Tamron 24-70 (and all other Sigma/Tamrons) could perform better af-wise on the 5d3 than the original Canon 24-70 (and many other Canon lenses)?!
The question now is if the disabled double-cross sensors in "group b" are (just) a marketing ploy, or if there is a real technical reason too like much distortion and/or sharpness falloff. Seeing the Tamron's reviews, it could very well be that the top and bottom two double-cross sensors are enabled, but won't be precise so it's a draw to the Canon mk1 or even worse.
It is hilarious that most 5d3 reviewers don't even seem to bother to read the manual and tend to say that only the max. open aperture is the decisive data that enables/disables the 1dx/5d3's af points. So with this even more complicated issue, it might be next to impossible to tell how 3rd party lenses (f2.8 or other) really perform with the new af system.