I am sure that such a sensor is technologically possible today, from Sony or Panasonic for example: the photosites would be about four times as large (in area) as those in recent phone sensors, and DR and noise levels would be fine for many uses. Also, in cases where the DR or SNR is inadequate, downsampling to a lower pixel count gets you close to the levels that you would get from a sensor of that lower photosite count, so those worries are not as great as some people think. (Selective resolution reduction as with noise reduction processing compares even better fro IQ against lower-resolution sensors.)
I even think it likely that Panasonic will offer a sensor of about 50MP (8192x6144) someday, to follow the 8K video trend.
But for now, the benefits of even more resolution are probably limited to a few extreme cases (like people who print huge, and those who are obsessed with "screen-sniffing": zooming on in little pieces of their images). So for now, the benefits for 4/3" format are probably outweighed by the disadvantages, making it not a good business decision to increase pixel counts substantially. Even discounting worries about DR and noise, the disadvantages of a higher pixel count include lower frame rates, slower buffer clearing after bursts, slower file transfer and processing, more storage needed, more heat production (particularly when recording video), shorter battery life, and so on.
The advantages might increase enough to make it worthwhile if and when:
- enough people are viewing images big screens with resolution beyond 5K (20MP 4/3" sensors are about 5184 x 3888; 5K screens are 5120‑by‑2880.)
- 8K video becomes a significant market.