What?! They're different tests. Someone who scores a 30 didn't compete with same testers as someone who scored a 508. They are different tests so the best you can do is to account for the overall percentiles. Someone who scored a 30 scored better than X% of testers on their test, and someone who scored a 508 scored better than X% of testers on theirs, which allows adcoms to "compare" tests without being skewed by the differences between them.
I think you're contradicting yourself by arguing that overall percentile ignores both the similarities and the differences...?
Someone who scored a 30 didn't compete with the same testers as someone else who scored a 30 on a different test date. I disagree that the best you can do is account for the overall percentiles.