Compare native strategies on historical ETF, FX, perpetual and dated futures inputs #187

The small workloads in #185 check that the adapters can run. They do not show whether the current Nautilus engine and the three strategy paths through it deliver the same decisions on historical data, or whether Backtest and pysystemtrade agree on a dated futures trend and roll.

Add optional workloads that check the following before recording timings:

• ETF and USD-quoted FX EMA decisions in Backtest, Nautilus with a Python strategy, Nautilus with registered Rust indicators and Nautilus's bundled Rust strategy.
• Long and short perpetual funding payments on the same eight-hour series, with both account balances reconciled.
• A dated SP500 and GOLD trend using native pysystemtrade forecasts and accounting, Backtest's futures roll and variation cash, and Nautilus's dated contract execution. Compare marked equity while retaining the different cash accounting methods.

Bind reports to the inputs, environments and code used. Keep these experiments separate from the release-qualified comparison evidence. No runtime compatibility change.

AI assistance: Codex investigated the APIs and prepared the benchmark implementations, checks and this issue.