fix: keep report LLM, memory-after-close, and evidence/stats projections

[munos1233]

Summary

• Report generation keeps the live LLM path: timeouts no longer fall back to templates; proxy/API timeouts match the longer report budget.
• Close paths spawn MemoryAgent after CLOSED without blocking the parent Celery/request budget; SuperAgent skips that pass when the graph already owns it.
• Effect verification rate and writeback envelopes only count immediate response/rollback entity-effect actions; timeline evidence expands sanitized observation_fields instead of repeating the node sentence or exposing raw_data.

Test plan

• Backend: report timeout / no-template tests, memory-after-close tests, writeback projection tests, stats effect-rate tests, evidence safe projection tests
• Frontend: EventDetail closed-state memory poll, StorylineTimeline observation fields, apiClient timeout
• Do not merge eval-matrix artifacts or local docs/仓库运作说明.md

Made with Cursor

[cursor]

Bugbot couldn't run - usage limit reached

Bugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit.

A user or team admin can review and increase usage limits in the Cursor dashboard.

(requestId: serverGenReqId_e6eff5cb-3789-4055-ad98-98beb288cd45)

[cursor]

Left a non-blocking comment and assigned reviewers. Cursor Bugbot skipped after a usage limit and did not complete a review, so this PR is not approved and needs human review.

  

_{Sent by Cursor Approval Agent: Pull Request Router and Approver}