You can obviously not interpret lenstip measurement , lenstip figures first and then photozone . Read Imagin-Resource test where they thought the lens was faulty. se Turners report.
Let's summarize. You say 50/1.4 has better resolution. DigiAngel's bicycle photos above (thank you!) show us the reality.
You say 50/1.4 is sharper overall and you cite the Lenstip tests. The Lenstip tests show that it depends on the aperture. For example, if you look at the centers (red), the 50/1.4 is better at f/1.4, and the 50/1.2 is better at f/2, and both lenses are equal at f/2.8, and the 50/1.4 is better at f/4. This back and forth lead is similar to what was found by LensRentals. It seems you took most of the blue (edge) dots out of the 50/1.2 chart from Lenstip, but I presume the blue dots mean the same edge point on both charts. There again, which is better depends on the aperture.
You asked me to look at Imaging Resource but your link is to SLRGear (?), so I clipped SLRGear's blur charts which show the 50/1.2 as distinctly better at f/1.4. The lower, darker and flatter the blur chart, the better the resolution, and the 50/1.2 lens definitely shows a better blur chart at f/1.4 and f/2; by f/2.8 they are comparable. SLRGear was disappointed with the 50/1.2 lens based on its price, but they did say (in their "Tanner Report") that "It's blur profile is somewhat better than that of the Canon 50mm f/1.4 wide open". Looking at the blur charts, there is no question about that.
You link to The Digital Picture (TDP), but their charts show that which is better depends on where you look and which aperture. For example, at f/1.8, the 50/1.2 shows a better center and mid-frame, but a worse corner. At f/1.8, I would rather shoot the 50/1.2 lens than the 50/1.4 lens. TDP's results supports this.
You dismissed the LensRentals test results as "wrong". Well, just dismissing test results doesn't help your argument about the "myth", especially when their results are similar to those on Lenstip, TDP and SLRGear.
I'll grant you that Photozone's test shows the 50/1.4 as much better than the 50/1.2; that stands out as the anomaly among all of these sources. Of all the sources you cited, this one supports your point.