I previously stated as much that I would have no issue with that. The main issue I think I had, and that you also rightly pointed out, was the lack of grouping/pattern notation. "2 COAXIAL HOLES" in your latest example would satisfy that.
I would agree that making A and B a commom datum feature would be valid, and mean the same thing as the current drawing. The presence of the 2 COAXIAL HOLES makes a difference, as it is a grouping mechanism that makes the position tolerance apply to both features "simultaneously" and control their mutual coaxiality. The OP drawing did not have this, it only had multiple leader lines, which is not a grouping mechanism.
It would be more interesting if the two features did not have the same size tolerance and were referenced RMB ;^).
I would say that is an YOYO example (to rephrase Don Day's statement)
YOYO = you are on your own (meaning you have to explain it to the end user because the standard is no help here)