I see a lot of people recommending to set the Sony zebras to anywhere between 105% and 109%. Specifically because of this characteristic.
Considering that exposure compensation is in 1/3 stop increments and I’m not a pro, the default behaviors for the camera and darktable are meeting my needs.
I use exposure bracketing rather than zebras. I quickly get three shots in a fraction of a second and on the computer in DT I can look at the raw clipping indicator. A blue sky might be clipped in the blue channel only but the clouds are rendered by the green and red channel and DT can handle this sort of clipping without problems in my experience. Now if all three RGB channels are clipped that is a problem that DT can’t resolve. I suspect the camera’s zebras are incapable of this level of information.
Memory cards are inexpensive so this approach of mine is not a burden, but it can have the advantage of three different exposures but also three different moments in time which might save a portrait shot if a person blinks or a wildlife shot if the animal turns away. At worst I end up deleting 66% of my shots later in the editing process.
What is the step in EV between bracketed shots?
On my older cameras I used +/- 1EV whereas on my Canon R7 I use +/- 1.67 EV because it has a more dynamic range of capture. I could possibly even push this further but most problems in lighting are covered by this range.
If photographing a backlit subject like a bird I would prefer to do 0.0EV, +1.67EV and +3.3EV. That would only be possible if I had time to adjust the exposure compensation dial to suit the subject. But when I just grab the camera and click quickly to capture a shot the normal bracketing covers my butt and one of the shots will work well. I do a lot of travel photography and sometimes I have to take a shot with very short reaction time.