Document in great detail where constraints for data leafs is authoritatively defined

Marking this as Critical as blocks 📌 [PP-1] Add support for matching SDC prop shape: {type: string, enum: …} Postponed which is critical

1. What does "all … are not found" mean? 🤔 Do you mean "not all … are found"?
2. This is indeed a tricky area, which is why there's this:
   

       // @todo Field item-level indirect vs direct constraints should not override each other. Investigate in Drupal core, this seems to be an oversight?
       // Field item-level constraints override property-level constraints.
   

   … because AFAICT Drupal core somehow does not (yet) provide infrastructure for this 🤯

3. RE: points in issue summary:
   1. Correct, "always using the first element" is what I discovered (during a deep debugging session) Drupal core always does … but now I wish I had added a comment pointing to where in Drupal core that is happening 😬
   2. Excellent remark — I didn't discover that edge case at all, but you're right that it's absolutely possible. We should at minimum add an assert() that verifies there's an empty intersection, and when we find the first non-empty case, we should match Drupal core's existing behavior. Ideally, Drupal core should provide an API for this …
   3. I'm fine with that simplification — I see that FieldItemBase::getConstraints() also promises this, so I have no idea why I did that. Probably defensive programming? 🤷‍♂️

If after this docs + logic improvement, 📌 [PP-1] Add support for matching SDC prop shape: {type: string, enum: …} Postponed is still blocked, please open a new issue with a failing test case to show what the precise blocker is. This represents a step forward, so I'm going ahead and am merging this.

Automatically closed - issue fixed for 2 weeks with no activity.