Which contact makes our drilled angle bracket rock in the UR5e nest?

Base and side are the intended locators. Both samples touch an extra end block before settling on the base.

Take that contact back to the designer with the actual accepted sample range. An end block that wasn't meant to locate can still do a very effective job of holding a part off the base. More clamp could hide it.

Was the end block added after the locating scheme was drawn? Looks like a helpful extra that got promoted without asking.

Yes. Added as a loading guide. Designer confirmed it interferes with the accepted end-length range and specified relief.

Does that preserve the intended base and side contacts?

It does. Replacement guide is made; bench checks with the identified sample range now seat on those contacts.

Can the usual loader tell when the bracket is properly seated, without whoever designed it pointing at the gap?

Loader and cover both demonstrated it. Same seating result after repeated placement. Thanks, Maya, asking where it touched found the unintended locator.

Keep the revised guide identity on the nest record. The old one may still look like a perfectly serviceable spare. This settles the bench-seating fault; the planned robot-loading trial has its own checks.