Split train and tool simulator modules; mastery curriculum and grader workflow nudge.

Harden notebook vLLM compatibility checks

Rebrand: drop 'Support Ops' user-facing text; add DriftShield* aliases; tighten README

Cleanup: drop legacy support-ops cruft now that DriftShield is the only collection

DriftShield Phase A: D1 collection, 4 runtime-failure tasks, ops.get_recommendation, recoverable schema drift

feat: GRPO training stack (Qwen3 + Unsloth opt-in, curriculum, eval, audit)

clean rewrite: shorter vars, no comments, same logic