[Klaus]

Just think how contrast-detection AF works - it tries to detect the maximum edge contrast pixel level.
It literally reads out the main sensor.

RSAs are focus shifts when stopping down. They do ALWAYS occur - the level can range from negligible to very substantial.

If the camera stops down prior to engaging the CD-AF, RSAs are simply irrelevant. There is no bracketing necessary (I suppose you meant bracketing and not stacking? Stacking is really more meant to increase the DoF in an image).

In theory, a DSLR could use bracketing around the PD-AF point - and do a contrast-detect post-analysis to select the best-focused image.
However, bracketing requires time - time you may not have. Bracketing would also introduce a wobbly viewfinder image.

Funnily, the latter has been criticized in Panasonic mirrorless cameras which use CD-AF only rather than a combo of PD & CD-AF (like Sony/Canon/Olympus/Fuji/Nikon) where this wobbling is not detectable.

That all being said - it's all a question of how holy you want to have it. An occasional misfocus is usually not the end of the world.
I do, however, appreciate "animal AF" for instance.