feat: add test-window coverage reporting

[AKKI0511]

Summary

• add per-symbol test-window coverage metadata and persist JSON/CSV reports during training and backtesting
• surface coverage summaries in CLI commands without altering existing result payloads
• expand time-aware split tests to verify coverage scenarios and adjust pipeline validation expectations

Testing

• make format
• make lint
• make test
• poetry run pre-commit run --all-files

---

Codex Task

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

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 run_pipeline return type

run_pipeline now returns a tuple (results, coverage_info) (lines 388-391) instead of the documented results dict, so any existing programmatic caller treating the return as a mapping—for example run_pipeline(...).keys() or results["AAPL"] as shown in the docstring—will now receive a tuple and hit type errors. The commit summary also claims to avoid altering result payloads; adding coverage metadata should be done in a backward-compatible way or it breaks current consumers.

Useful? React with 👍 / 👎.