[codex] Support time-varying report risk-free rates

[luuhongyii]

Summary

• allow report paths to accept a time-varying risk-free rate series
• subtract Series risk-free rates from DataFrame returns by date index
• display time-varying report RF values as an average in report parameters and metrics
• keep single-value aggregate metrics from expanding when RF is a pandas object

Root cause

reports.html() and reports.metrics() accepted rf as if it were always a scalar. Passing a pandas Series triggered ambiguous truth-value checks and scalar formatting failures. The underlying excess-return helper also subtracted a Series from a DataFrame along columns instead of rows, which broke date-aligned RF calculations inside reports.

Fixes #521.

Testing

• PYTHONUTF8=1 python -m pytest -q (129 passed)