There's one key point you're still missing here. If there really were
such a variance in shooting position, coupled with focusing errors,
then it would lead to clear inconsistency in the resolution results
for the 10Mp Four Thirds cameras. But in contrast, they all give much
the same result. So let's consider this in a bit more detail.
If that EXIF distance data is an accurate measure of camera is
subject distance, then the inescapable conclusion would be that the
E-510 result must be well out of focus, or the magnification of the
shot must be different. In either case, the resolution result would
be significantly different from the other 10 Mp Four Thirds cameras.
But it isn't.
Also, you need to appreciate that these res charts are shot using
manual focus bracketing. It makes no sense to go to the effort of
setting up the camera and shooting the chart, then finding it's OOF.
This effectively means that at least one (and in practice, probably
several) in-focus shot is always obtained, and subsequently used.
What you've actually discovered is a concept familiar to anyone with
a scientific or technical background, which is that even with the
best will in the world, there will always be some variance between
experimentally tested data, and the job of the investigator is to
identify and minimise contributory factors. Naturally, we make every
effort to do so.
So ultimately, the situation is that we have one piece of data (the
EXIF distance) which looks wrong, but in fact very consistent
resolution test results across the 10Mp Four Thirds cameras. And as
yet we have no idea about the accuracy and reproducibility of that
EXIF data.
So your complaint about shooting distance is interesting, but
currently has to be filed as 'not proven' at best.
