Because that's where headspace is measured on a rimmed case such as the .303. There are 2 separate & distinct, but interconnected problems here & to get the best results we need to fix any problem with both of them.
Let me try & explain differently. Take a look at these 4 "cases"
Top is the whole ding, dang doodle loaded round, lets assume its all within spec & works as we're going to use it a a reference.
It breaks into 2 sections, the rim, which is also the headspace, & the body, which is the chamber dimensions. (we're assuming a fired neck resized case for the sake of argument.)
Now lets have some excessive headspace, but with a normal chamber. What happens?
Well the normal body is pushed forwards by the excessive headspace, so to fix the problem we need to fix the cause of it, the headspace. If we fix the headspace by changing the (perfectly good) chamber dimensions we've just compounded the error.
Lets look at the other problem, "generous chamber".
Its the wrong length & size in the body of the case, but the rim (where we headspace) is fine. Can we fix this with headspacing? Nope, this is where we fix the chamber dimensions to be correct.
So what's the "right" way?
Step 1: Fix any headspace issues by changing the bolt head. Why? well, because if we can't hold the rim in place everything else is moving with it causing all sorts of false readings.
Step 2: Now we
KNOW the headspace is correct & the round is immobilized we can see (for the first time) how much of the problem is headspace & the case flopping back & forth & how much is an over sized chamber.
OK, so what does this do for us?
It works the brass less.
It is more consistent, because its always in the same place (don't forget the bullet is moving back & forth as well, changing the jump to the lands)!
It fixes the 2 problems independently, actually fixing the problems instead of coming up with a "fix" that just hides them.