It is fair to say I think, that the final image is a result of the entire imaging chain. I will go out on a limb a bit to also assert that with digital, "great glass" assuming that could be defined
is significantly less important than with film. Remember that the legendary lens makers earned their reputations during the film age and have to some degree benefited from that long ago earned halo.
Even in film days, the look of a lens could be varied somewhat by the development technique, "sharper" developers would essentially create what we now refer to as "sharpening artifacts" due to local development agent exhaustion at image edges.
Today, sometimes it is one thing in the chain that is the obvious weak link. AA filters come to mind, but for some applications moire is a plague (wedding shooters raise your hand), so they have a good application when well balanced.
There are some digital artifacts that cannot be resolved no matter how complicated, by lens design (cosine law effects, for example).
Once software becomes part of that image processing chain, then there are all sorts of possibilities on improving and correcting what the combination of lens and sensor deliver, and that is always true despite the degree of lens and sensor "perfection" at least to-date. Design of all things is a balance of costs, goals, and imperfections, a compromise so to speak. All the "no compromise" rhetoric in the world can't change this.
If one uses jpeg camera files, the processing is limited to that recipe installed in the camera, and often that is limited by the limited processing power available to deliver the acceptable frames per second. A computer based raw processor mostly removes that constraint and gives wonderful controls.
The difference between a lens that provides 60% and 80% central contrast at say 30-40lpm is covered easily by just a small slider adjustment in post.
So, and to the extreme, if your finished product is the print, there are further complications that make some of the differences in lenses and sensors even less significant; each paper and printer and size combination has its own "best" processing.
The important invarient characteristics that are important in image making has a little more to do with the DR, linearity, and noise characteristics of the sensor and the choice made by the sensor maker of the shapes of the response curves of the filters used for each bayer primary.
So I mostly agree that the result is the significant point, and that there are a number of choices that can yield equivalent results in the finished product.
Those manufacturers that have a rich software capability well matched with their camera and lens system are less disadvantaged than those that do not.
-bob