fix(server): classify Claude V2 structured terminal failures

[saphid]

Claude structured failures can carry success-shaped SDK results. Current V2 already classifies these failures; preserve that upstream behavior and its authentication/rate-limit hints, retain the terminal reason as the failure code, and prevent structured error text becoming a synthetic assistant reply.

Rebuilt on V2 7dd9873ed with test-only prerequisite #10865 (merge first). Current head 6f221e835; remaining topic is two files, 69 additions and 4 deletions.

Validation: 132 focused Claude adapter and persistence tests pass, plus server typecheck, targeted lint and formatting. New CI pending. Direct Claude Opus 5 high review remains unavailable: launcher exited 1 on expired OAuth before inference; no Claude review occurred.

Targets V2 #2829. Updated with GPT-6 in Codex/T3.

[cursor]

Cursor Bugbot has reviewed your changes using default effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 34fb3afa45c930d74626a510d7c77a10af98e98f. Configure here.}

[macroscopeapp]

Approvability

Verdict: Approved at 6f221e8

Macroscope's review found this PR approvable — This is a contained Claude adapter fix that correctly surfaces structured terminal failures as failures instead of synthetic assistant messages, with targeted regression coverage for the supported failure reasons. The remaining change only updates a persistence test assertion and does not alter production persistence behavior.

^{You can add or adjust custom eligibility rules. Learn more.}

[cursor]

Bugbot is paused — on-demand spend limit reached

Bugbot uses usage-based billing for this team and has hit its on-demand spend limit.

A team admin can raise the spend limit in the Cursor dashboard, or wait for the next billing cycle to continue.

Dismissing prior approval to re-evaluate 6f221e8