Really interesting.
I t would be even more useful if you could share the raw.
Anyhow, modern sensors can have some various implementations of the electronic shutter, especially those realized through a stacked CMOS process.
One possible explanation for the above, I think, is that some sensors have a way to read the rows and/or the columns not sequentially, but in a sort of alternated way (e.g. 1-8-2-9-3-10-…). The reasons for these are essentially technical workarounds to improve the speed or size limitations, like the fact that a conversion unit (e.g. an ADC) may not be fit in the width of a pixel, hence the columns are not read sequentially.
Mine above is just a possible explanation, based on direct knowledge of some actual sensor implementation I have, but it’s not the only possible one. I haven’t looked deeply in the exif nor in any metadata, just looked at a 1:1 view of the picture.
It’s really cool to see some of these technical-low-level details showing up in some actual shots. Thanks for sharing.