Refer to https://docs.google.com/document/d/1gfwVdVdN54OeWzl_fiCP8ZY-Taz1lsrnTcFI1zNWNhU/edit?usp=sha ring Apex needs to implement Planned change 4 and Technical requirement 4. Match each state problem to the best control. Which mapping is correct?

The correct mapping addresses each state problem with its most direct control. Problem 1, "resume-agent-state loads .github/agent-state/304.json, but the active issue is 318," is resolved by requiring the stored issue number to match the active issue (1-B), preventing an agent from working on the wrong issue. Problem 2, "The stored state references pull request 701, but the active pull request is 742," is solved by requiring the stored PR number to match the active pull request (2-C), ensuring consistency with the current development branch. Problem 3, "The stored state commit differs from the current PR head commit," is mitigated by requiring the stored state SHA to match the current PR head (3-D), preventing work on outdated code. Finally, Problem 4, "Copilot CLI uses a copied local summary that includes a rejected Terraform apply approach," is best addressed by excluding rejected local summaries from the resume context (4-A), ensuring the agent doesn't re-propose discarded solutions. The other options misalign these direct relationships, leading to incorrect or inefficient solutions.