Unfortunately, DXO assigns a numeric value to their rating which does not tell you all the things that might be important to you, and is of little or limited value to me.
As with synthetic tests of this nature this is exact problem. They test with many possibly useful, pre-defined parameters and try to take it down to a numbers game. However what you are shooting is not their pre-defined shots. I also find that many great lenses just are not the right tool for a particular shot. A test like this is kinda like any other review, they can often get you into a ballpark idea of how things can compare but they are not the end-all be-all when evaluating a lens.
FWIW, I have the Tamron SP17-50mm VC and it is a very nice lens for the money and I feel that in many of the environments I shoot in it is nearly equal to some L glass, but there are other factors to consider such as weather sealing, ergonomics, etc. I have been using Tamron CCTV lenses for a very long time and they are quite good and I feel that the 17-50 is nearly the same quality as those, but the price difference is also a factor of many things beyond just quality of glass and other components. there is the quality and frequency of QA testing, brand recognition, marketing, in-field testing, and more that factors into pricing and Canon tends to be a bit better tested with higher QA and the L glass offers other benefits like the weather sealing that can be invaluable to some.