feat: harden SOXL RSI2 compounding acceptance

[Pigbibi]

Implements the frozen T4/S5 acceptance hardening for research-only SOXL RSI(2) evidence.\n\nValidation: focused pytest, cached Ruff, compileall, and git diff --check.

[github-actions]

🤖 Codex PR Review

🚫 Merge blocked: 1 serious issue(s) found in high-risk files

🚫 Blocking Issues

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

1. 🟠 [HIGH] Reliability in src/us_equity_strategies/research/soxl_core_optimization.py

This changes the outcome value on the success path from CHARACTERIZATION_CANDIDATE_FOUND to COMPOUNDING_CANDIDATE_FOUND while keeping schema at qsl.research.soxl_rsi2_mean_reversion.v1. Any current v1 consumer keyed to the old enum will now misclassify or reject successful results on this reachable branch, which is a wire-compatibility break under an unchanged contract. (line 1065)

Suggestion: Keep the existing success enum for the v1 payload, or bump the schema/version and update all readers/readback code to accept the new enum before emitting it.

ℹ️ Other Findings

1. 🟡 [MEDIUM] Reliability in src/us_equity_strategies/research/soxl_core_optimization.py

The new winner is None branch emits the same v1 schema with locked_winner: null and an evidence_gates object reduced to a single key. This PR treats that branch as a normal evidence_valid=True result, so existing v1 readers that deserialize locked_winner as a string or expect the full gate map can now fail or silently drop diagnostics on that path. (line 1007)

Suggestion: Preserve the existing field types and key set for the v1 contract on the no-winner branch, or publish this shape change under a new schema version and migrate consumers explicitly.

---

Review by Codex PR Review bot • PR

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 7420f39d0e

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: df47e8fcd7

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]

Include the prior close in the SOXX drawdown path

selected_soxx correctly starts at raw_start - 1 for the new return/CAGR comparisons, but metrics is built with _volatility_metrics_with_soxx, whose SOXX drawdown still starts at raw_start. Thus, when SOXX falls on the first session of a validation/test/holdout window, the strategy drawdown includes that session relative to its prior-session starting equity while soxx_close_path_max_drawdown omits it. The new final-holdout drawdown gate can therefore incorrectly reject a candidate (or report an incorrect benchmark-relative drawdown); compute the benchmark drawdown over the same prior-close-to-final-close path.

Useful? React with 👍 / 👎.