Three ideas I bring:
1) Isn't optimization what Leica does with their lenses? I used them mostly wide open. Is that different from SWEET SPOT?, which in many lenses is about f8?
2) Let me see if I understand focus shift:
If I have my camera with the Zeiss 50/15 on a tripod and focus perfectly on a model's eyes wide open (f1.5) and shoot, I should get the eyes on focus.
Then, if I touch only the lens aperture ring and move it to f2 (without touching the focus ring), will the eyes be in focus? If yes, there is no focus shift. If the eyes are not in focus, is there a focus shift, correct?
Please confirm.
Your theoretical example is correct. That would be an example of focus shift, in a lens design which has inherent focus shift, such as the Zeiss 1,5/50 of any vintage.
HOWEVER, the modern Zeiss C-Sonnar 1,5/50 as it came from the factory originally, would not work that way. It came from the factory such that the eyes in your example would be in focus with the aperture set at f2.8, not F1.5. If you set the aperture at f1.5, following the method you outlined, and focused on the model's eyes, at minimum focus distance of 1 meter, in the resulting negative it would be the tip of the model's nose, more or less, which would be in focus. You won't have missed focus by much, but it would be off.
If she were 15 feet away, and you focused on her eyes, both her eyes and nose, and more would be in focus, and you'd not be aware of any "problem" with focus shift.
If you had one of the C-Sonnars that had been modified to be spot on at f1.5, instead of f2.8, the results would be different accordingly, but it's easy to compensate either way in those cases (rare for some of us) where you are trying to nail critical focus at minimum focus distance wide open.
If the model and photographer both move 1 cm, either in or out, after focussing, then all this goes completely out the window anyway. And, of course, that never happens.