Compare this:
A mechanical engineer draws a traditional mechanical drawing of a metal sheet with 42 holes in it. The drawing is aimed for the person who will mark the midpoints of the holes on the workpiece then drill the holes. The mechanical engineer will not even consider anything else than drawing the measurements of the hole centers; even though they are "imaginary" points in the final piece, they are what the workers need.
Now consider component footprint drawing.
A mechanical engineer draws a component footprint with 42 square pads in it. The drawing is aimed for the person who will enter the midpoints of the pads into their CAD. The mechanical engineer will not even consider anything else than drawing the measurements of the pad centers---- oh wait!
So it's definitely not a rule or tradition of "mechanical engineering". It's a tradition of footprint drawings, and completely arbitrary.
One simply cannot compare round holes with square pads. Their shape alone is quite different let alone conventions for dimensioning them.
There are many ways of placing a pad and defining its location, sometimes having an offset from centre is the best option, sometimes not, maybe even its an arbitrary shape... a table of recommended xy values then becomes a less intuitive way of presenting the information. Its obviously a bigger concern for irregular components, but nonetheless, it happens, and until we have one EDA package to rule them all, there remains many different options for "entering" the footprint information, yet theres far fewer ways of measuring and verifying it.