fix: wake pending inputs after environment initialization

[sam2tom]

Environment initialization could commit after its earlier capability hint had already been consumed. Pending inputs then waited for the next candidate scan even though initialization was ready. Wake the existing scheduler only after the completion transaction succeeds; keep its ownership, capacity, Session exclusion and polling checks.

Add bounded stage timing for preparation send/wait, initialization commit and nonempty lifecycle reconciliation, plus admission wake-source counters. This does not reorder lifecycle observations, alter Runtime wire or change database schema. Operator documentation describes the overlapping intervals and measurement limits.

Validation:

• The same candidate-selection regression fails against the original initialization code and passes with this change; failed commits do not wake scheduling.
• Independent execution race suite: 334 passed, no skips or failures; focused concurrency tests repeated 20 times passed.
• Naming, documentation translation and diff checks passed.

No deployment or new production model calls. The previous cold admission sample remains 15.280 seconds; local regression evidence does not establish production latency improvement.

Independent full-diff review found no actionable P1/P2. CI run 38138612328 completed with 12 successful checks and 7 path-based skips on 3d2c3987ff25650bfb441e1689a292b8f5ae6cd2.

Sequence diagrams and independent QA evidence: SandBase #1127.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-11T12:19:01.661365Z	3d2c398	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.