It's so much more complicated than what's being discussed here. Pixel size and pitch are but one of many variables that can affect imaging performance. The old "rules" don't so much apply when you have generations of technological difference (high gain vs dual gain etc) and new designs like BMD's that have tiny pixels but also W filters that are a lot more sensitive than filtered pixels.
The main issue with 135 format is LEGACY film mounts. EF was designed as a FILM ERA lens mount.
Film era lenses tend to have very poor fall off performance. The corners are darker than the middle of the lens.
One of the main design goals of the MFT mount was to make a highly telecentric mount, meaning that lenses could be optimised more easily to not suffer from darker corners (and also other optical distortions)
This is called shading, or falloff.
https://www.edmundoptics.com/knowledge- ... gn-topics/Every sensor typically has a micro-lens on each photo site and these can also be very difficult when interacting with non-telectrinc lenses. Think about CA and then think about how every photo site has it's own lens in-between your taking lens and the photo site itself.
This is called sensor shading. It's like lens shading, but a secondary step at the sensor level.
For this that remember the infamous "magenta corners", this was to do with sensor shading.
Many modern lens mounts are designed for telecentric lens designs. Think LPL, MFT, R, L, E, Z mounts.
They are all optimised to be telecentric and have no requirement to work with legacy film era lenses which are often very far from teelcentric designs.
These choices affect the image far more than the size of the pixels.
JB