Intriguing observations. I find it amazing that the 14-54 which obviously
has a limit of 4-8MP of effective resolution shows far more resolving
ability on the E-3 than my E-1. If the lens truly was this limited, we
would see very little difference in image quality between the 5MP sensor and
the 10MP sensor.
As to diffraction limits, there is theory of diffraction and then actual
practice of diffraction. The airy disk isn't a "bubble" with all or nothing
blurring, but it is a general fall-off with a penumbra. As such, diffraction
limits are not hard barriers, but general loss of clarity in the extremes of
resolution. The more diffraction you throw at it, the wider the penumbra
and eventually the umbra of the airy disk covers enough neighboring pixels
to finally limit the maximum effective resolution.
Since the diffraction limits are somewhat soft, a very slight sub-one pixel
sharpening will usually counteract and invert the penumbra hence recovering
the details otherwise obsured or diminished by the diffraction.
Themed Olympus Photo Exhibition: http://www.tope.nl/