Tbg, we seem to be in a circular argument.
First let's state a hopefully agreed fact: whether the Muratas, Townshends or another supertweeter, their frequency bandwidth starts at some point. For example, you use the figure of 15KHz for the Muratas.
Not every speaker on the market is polite enough to stop at that selected point. Some fall short before. Some run past that frequency. Perhaps a few stop where the supertweeter expects them to.
Your position seems to be: no matter what the main speaker, a supertweeter always makes things better.
Some of us are less credulous of that position. In Martin Colloms case, he is one of the most respected and experienced voices in the world of speaker design and evaluation. I, and many others, give his views great credence. A supertweeter, if used, deserves the same effort toward integration as any other driver. Otherwise the effectiveness of the results will be haphazard.
I guess we just need to agree to disagree on this point.