feat: legacy-safe result/store timing codec

[Pigbibi]

Scope

R1S2 redesign v2 PR A only: legacy-safe timing metadata and store codec.

• BacktestResult timing/identity metadata is keyword-only and positional-compatible.
• v1 records remain readable; v2 timing identity is explicit and canonical profile IDs are serialized.
• Default and explicit None remain legacy-only; LEGACY_EXECUTION_TIMING is explicit legacy-only; ANY_EXECUTION_TIMING is the only cross-timing query.
• Timing keys are collision-free; store write revision is the tie-breaker with deterministic legacy fallback.

Spec: /Users/lisiyi/Documents/Codex/2026-07-14/ba-2/outputs/soxl_tqqq_result_store_redesign_v2_2026-07-14.md (PR A)
Commit: db366f157c1c59a55cbc739d87cc23c755152b72

Non-goals

No caller migration, orchestrator routing/copy paths, capability wiring, persist_mode, ephemeral/exporter, broker/live/account, or frozen PR #248/#251 changes.

Validation

601 full tests passed, 1 skipped; 62 focused/related passed; ruff, compileall, diff-check passed.

[github-actions]

🤖 Codex PR Review

🚫 Merge blocked: 2 serious issue(s) found in high-risk files

⚖️ Codex Review Arbitration

🚫 block: 第一个阻塞点仍然成立：known_profile_aliases() 只展开规范名、注册别名及其 lower()/upper() 变体，load_latest_backtest() 也只按这些精确前缀做枚举，所以历史 v1 结果若实际落在 backtest/.../SoXl_SoXx_TrEnD_InCoMe/ 这类任意混合大小写目录下，按规范名 SOXL 查询仍不会扫描到。canonical_profile_id 只在新写入时持久化，不能修复既有 legacy 路径；新增测试也只证明“用同样的混合大小写去查能读到”，没有证明“规范查询能命中任意历史大小写”。不存在与 prior finding 的 contract conflict：当前第二条关于写入顺序的建议与 prior finding 同向，只是它本身已被新增的 test_any_timing_is_stable_independent_of_save_order 这类测试所反向约束，但这不影响第一条已被源码直接证实为真。

🚫 Blocking Issues

These issues must be fixed before this PR can be merged:

1. 🟠 [HIGH] Bug in src/quant_platform_kit/strategy_lifecycle/performance_store.py

load_latest_backtest() 只枚举规范 ID、注册别名以及这些别名的全小写/全大写前缀。v1 时代回测文件路径是直接用原始 strategy_profile 落盘的，路径在本地文件系统和对象存储里都是大小写敏感的，因此像 SoXl_SoXx_TrEnD_InCoMe 这类历史混合大小写前缀，后续如果按规范名 SOXL 查询将不会被扫描到。结果是历史记录明明存在却返回 None，与“v1 records remain readable”不一致。 (line 294)

Suggestion: 查询规范 profile 时不要只靠有限别名展开；要么基于持久化的 canonical_profile_id 建索引，要么在读取时做大小写无关的前缀发现/迁移，确保任意大小写的历史 profile 路径都能被规范查询命中。

2. 🟠 [HIGH] Logic in src/quant_platform_kit/strategy_lifecycle/performance_store.py

_backtest_sort_key() 现在用 (computed_at, param_version, result_identity_version, key) 排序，没有把写入顺序作为 tie-breaker。这样一来，只要两条记录共享相同的 computed_at 和 param_version，load_latest_backtest(..., execution_timing=ANY_EXECUTION_TIMING) 选出的就会是文件名字典序更大的那条，而不是最后写入的那条。由于 _backtest_key() 已把 timing 编进文件名，这会稳定地偏向某个 timing，导致“latest”读取到陈旧结果。 (line 650)

Suggestion: 在 save_backtest_result() 持久化单调递增的 store_write_revision（或精确写入时间），并在 _backtest_sort_key() 中优先用它做 tie-breaker；只有读取没有该字段的 legacy 记录时，再退回到确定性的 key 排序。

---

Review by Codex PR Review bot • PR

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: db366f157c

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".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: a8738cc9de

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".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Preserve timing when orchestrator persists results

When a runner returns a timed BacktestResult through the standard BacktestOrchestrator.run path, persist_result reconstructs the result in src/quant_platform_kit/strategy_lifecycle/backtest_orchestrator.py:86-115 but never copies this new execution_timing field. The default None then makes the saved record use the legacy t0_missing key, so load_latest_backtest(..., execution_timing="next_open") cannot find a timed run unless callers bypass the orchestrator and save directly.

Useful? React with 👍 / 👎.

[Pigbibi]

Contract arbitration note: the write-revision HIGH is not adopted. PR A source-of-truth defines ANY_EXECUTION_TIMING selection as deterministic and independent of save order; the tie-break is computed_at, param_version, result_identity_version, then stable key. The final tests exercise reversed save order and produce the same selected timing, so no wall-clock or monotonic revision is introduced.

[Pigbibi]

PERMANENT_FREEZE: arbitrary mixed-case legacy profile discovery requires an explicit prefix inventory/index/migration or narrowed compatibility contract; no further local patch. Preserve this PR as frozen evidence. The write-order HIGH remains rejected: ANY selection is deterministic and save-order-independent by approved contract/tests.