Thanks people for taking the time to do the test, it's very helpful! Looks like this is a rather isolated problem... (although D-Nice did report that his display had similar issues, which prompted me to do a broader search for affected displays)
Some people have asked me to provide JPEG files for use in TVs, I added them to the first post. They should be relatively immune to poor cabling or deinterlacing.
In general, I agree that test patterns are not necessarily representative of reality. For example, my TV blinks considerably when displaying certain pixel patterns, and clearly they never happen in normal programming so this is not an issue at all and nobody in their right mind should pay attention to such theoretical junk. But the test pattern I gave is not in that kind of league, because the underlying problem is really serious and has an impact on everyday viewing. In the case of the PZ81, this means washed out grays and "fluorescent" reds in dark areas of the picture, and bright whites contrasting with dark blues in bright areas of the picture. The reason for providing a test pattern is that you can't ask people to look for this issue and hope that the results will be reliable -- therefore the test pattern is a very convenient way for everybody to be able to objectively measure the behaviour of their TV so that we can map out the full extent of the problem. The pattern is just a means, not an end.