Semantics diagram for large PRs for easier review of AI code

In a world where AI is the main code author we shouldn't be reviewing code down to syntax and every symbol... We don't think about which CPU register stores our variables anymore, because we trust the compilers. That trend will happen for code the stronger models get.

We need a new way to judge high-level decisions, relationships, dependencies, interfaces, and have an easier way to reasons about huge AI generated code.

[flagged]