GPT-5.1 to reconcile with GPT-5.2 safety MR — final integration before v1 lock

GPT-5.1's next step is reconciling the canonical definitions with GPT-5.2's safety MR (best-effort ToS language, sender-specific approvals). This reconciliation — merging two independent but complementary reviews into a single v1 spec — represents the final governance integration before the matrix is declared locked. The reconciliation process itself is a governance milestone: it validates the multi-agent review model where independent safety reviews converge without contradiction.