I'm not certain what you used as your focus point. The hyperfocal distance of a 35mm focal length at f/8 is about 27'. If the camera focused on anything nearer then that would explain the issue.
Also, a lens doesn't necessarily have a "flat" focal plane. It's typically just slightly curved and a bit closer at the corners than it is in the center (although lens designers do try to compensate for this to create a nearly flat field.)
I'd be interested in knowing how those tree leaves in left edge of frame look as well as a how the center looks (it'd be easier to judge this using the original out-of-the-camera image.)
If you look at the meridonal performance at f/8 for your lens at the 'wide' end (visit this page and click the "more" link at the end of the description to expand the page to show the full description, sample images, and MTF test curves) you'll see that the lens does have quite a noticeable fall-off in it's ability to resolve fine detail at the edges and corners of the frame (in other words, some loss of detail resolving is normal.)
There are other factors that can certainly come into play such as camera motion of a hand-held camera. But your shutter speed was 1/640th at a 35mm focal length -- which is extremely good (you should have no difficulty hand holding a 35mm at 1/640th ... even if the lens didn't have image stabilization). So I doubt that's the issue (I suppose you could blur it with motion, but you'd almost have to be trying to move the camera while taking the shot.)
BTW, the blur I see seems to be along the radial direction (from center point radiating to the outer edges and corners). Such a blur would cause a poor score on the MTF curve for the "meridonal" direction. On the Canon MTF curve, you'd want to look at the blue dashed lines. There are two such lines... one thick, the other thin. The thicker line shows performance using the thicker/bolder test lines (with larger gaps between lines) on an MTF test chart. The thinner lines shows the performance using the finer lines (with narrower gaps) on the test chart. Basically the finer line shows how the lines can resolve finer levels of detail.
See this page for an explantion: http://learn.usa.canon.com/resources/articles/2013/reading_MTF_charts.shtml
The left edge of the graph shows lens performance in the very center of the lens (center axis of the image). As you move to the right side of the graph, it's showing the perfrmance some number of millimeters away from the center point (as measured on the camera sensor). An APS-C sensor camera has a diagonal measure of roughly 28mm... so the corner would be 14mm from center. The far right side of the graph shows the performance of the lens in the extreme corners.
The vertical direction measures quality at that distance. If the line is above the 0.6 mark then the lens is doing fairly good. If it's above the 0.8 mark then it's extremely good. If it's below the 0.6 level then it's fair to poor.
I can't tell if you have the "STM" version of this lens... that's the better version. The non-STM lens is extremely poor in the coenrs.
The non-STM version of the lens is good to about 7mm from center axis before it's ability to resolve fine detail drops out of the "good" category and into the "fair". At 10mm from center it really takes a nose-dive into the poor/marginal category.
The STM version does quite a bit better (but still not good enough to compete with Canon's top end lenses). The STM lens makes it to about 10mm from center axis before it drops out of the "good" category and while it does get into the "fair" category around the edges, at least it avoids the "poor" category (e.g. below 0.4 one graph).
You do have to be careful when staring at MTF charts. First... they don't translate well across completely different types of lenses. Second... lenses that can provide coverage for full-frame cameras have MTF charts where the horizonal axis is tested out to about 22mm from center axis (whereas an APS-C sensor camera using EF-S lenses only uses about the first 14mm... so if you were wondering how an EF 70-200mm f/2.8L IS USM II would perform on your camera, you would ignore everything past the 14mm point because your sensor isn't large enough to capture anything beyond that point.)
Also... MTF charts only provide insight into one particular attribute of a lens... and there are so many. Sometimes a lens that doesn't have what you'd think of as a fantastic looking MTF curve... produces gorgeous-looking images that are hard to beat. So it's not all about the MTF curve.
You might be wondering... how DO photographers get staggering corner-to-corner sharpness (maybe you've seen such an image.) These are often made by "stitching" images together. Divide the image into a grid and the center area of each image can be used to provide the detail for that grid-square. That's one technique (and there are others.)
Tim Campbell
5D III, 5D IV, 60Da
