@bfeng
doing something for twenty or one hundred years may sound as a reason, but it isnt, that is called tradition, reasons lay underneath
as i can see it, there are serious mechanical reasons to address that it is exactly the same to perform the adjustment at any of those ends
the contact between the eccentric and the tip of the valve is round/flat
between the cam and the lobe is round/round
in both cases it is linear contact
if you have a defect/deterioration, in any end, you will not solve it by measuring at the other end, there is no way to have the stems worn into the eccentrics, as the valves rotate with each movement, and the top valve is a very hard metal, yes you can make the eccentric surface flatter, but that has no effect in measuring the gap, maybe it will make that valve gap more difficult to fix as the eccentric is round no more
the only advantage i see is that when you try to pass the feeler gauge at the valve end you find more difficult to introduce it as the rod of the valve is a vertical element, while at the other end you slide the gauge over the cam lobe, not sure about the access to that point in each of the rockers, there are many solid aluminium walls around there
i can see one point that is unclear to me as when you do it at the valve you put pressure with the feeler gauge, the kind of pressure that you have to be sure is not compressing the valve spring or you will mess up the value, well when you make it the other end you have a lever to put that pressure on the spring maybe making easier to disturb the measure, maybe ?
i can understand that the books recommend not to do the cam end to avoid using a wrong value, being 0,25mm at the valve someone may not divide into 1,26 to make it at the cam
i try to understand the reasons