Upload & Sell: On
Using outer auto-focus points on an f/1.4 lens, and then shooting wide open, is a stern test. Problem is that the outer part of the frame is never as sharp as the center, and the focusing is done using that less-sharply-rendered part of the image, which probably makes your auto-focus system guess a bit. Then you shoot wide open, assuring the least possible depth of field, so no chance of correcting any focusing errors by recording added depth of field in the image you make.
When people tell me they get reliably perfect results doing it that way, I'm inwardly skeptical, but don't discount what they say to zero, because, hey, they know what they see and I don't. But it's a stern test. My 5d III does what I consider to be a superb job with the 35L wide open, including on the outer points. I use them without hesitation. But they will occasionally be a bit off, and the center points are a bit more reliable.
You say the weak points are spread out. Have you tested systematically enough to prove that some points are always bad, and others always good? That would be useful information.