feat: require exact Codex residency resume

[schickling-assistant]

Problem

A cold Codex runtime must resume its exact checkpointed native thread. A fresh-thread fallback would silently lose continuity.

Goal

Add a capability-complete Codex checkpoint and exact native-resume slice on top of the residency foundation.

Decisions

• Validate agent, runtime, generation, thread identity, and checkpoint integrity before spawn.
• Lower only the verified native resume selector.
• Reject authored selectors that could bypass the checkpoint.
• Refuse the launch when exact native resume is unavailable or stale.

Verification

• cargo test --lib codex_app_server::tests -- --nocapture
• Focused runtime-owner-lock rerun passes; the full module has one known Darwin suite-interference failure tracked by PR fix(codex-app-server): deflake darwin-parallel test races #318.
• cargo check --workspace passes on the complete stack.

Complexity

Checkpoint validation remains in the Codex driver boundary. The host policy added later calls this capability through the shared residency model.

Concerns

This PR prepares and validates native resume but does not yet schedule cold transitions.

Friction & bottlenecks

A pre-existing Darwin parallel test race can fail runtime_owner_lock_is_nonblocking_and_released_on_close; its isolated rerun passes.

Follow-ups

PR #492 adds the equivalent OMP capability.

References

Depends on PR #490. Part of stack #495. Related test deflake: PR #318.

Posted on behalf of @schickling

field	value
agent_identity	mbp2025.direct.omp.bn5hrz39
session	mbp2025.bn5hrz39
agent_persona	generalist
agent_supervisor	unavailable
agent_tool	OMP
agent_tool_version	18.1.7
agent_runtime	OMP 18.1.7
tooling_profile	dotfiles@4816a7a