Although avast could improve the cleaning capabilities, I'd rather always put the priority in detection first.
There are other tools for cleaning, and even better than that, users that are specialist on cleaning malware (with logic, knowledge and brain).
Those tools targeting cleaning aspects are a complement to avast detecting capabilities.
So, from my point of view, the test is just one additional parameter. I don't question how "valid" or "close to reality" the test is. It is just a parameter, and there is no "good" or "bad" parameter as an absolute concept.
I, personally, keep measuring avast (or any other security tool for that matter) according to my needs, my own experience with it, and as I mentioned I'd rather always have avast detecting first hand (so please keep improving, always, on that aspect) than cleaning (which comes later in the security steps). More important than one test, is what happens in "real life" during longer periods of time.