What in the world are you talking about? Nothing m43 was at all referenced in this article, nor was any compensation made for sensor differences. Are you so bent on trying to take pride in m43 optics that you are spinning one middling result from a competing system's lens into some sort of greater revelation that past results are now somehow irrelevant or flawed?
Newsflash: DXOMark tests lenses from widely disparate systems with vastly varying pixel counts and sensor sizes. They have an established protocol and won't stray from it just so you can feel that your holy lenses remain holy.
The SEL20F28 is a compromise lens derived from a worse-compromised lens, the SEL16F28. It was always about compactness, and never about achieving high optical standards. Other NEX lenses, as DXOMark has shown, can be much better.