Add multi asset backtest support with full single-asset backward compatibility

[TheNoobiCat]

Why

The change makes multi-asset portfolio simulation a first-class capability in backtesting.py while preserving existing single-asset behavior and API.

The goal is to let users run one backtest across multiple symbols with minimal friction (ideally only changing data from a DataFrame to a dict[symbol -> DataFrame]), and keep strategy ergonomics close to current single-asset usage.

What I changed

1) Backtest input now supports multi-asset data

• Backtest(...) now accepts:
  • pd.DataFrame (legacy, unchanged)
  • Mapping[str, pd.DataFrame] (preferred multi-asset API)
  • Sequence[pd.DataFrame] (fallback; auto-symbols)
• In multi-asset mode, all input DataFrames must have identical aligned indexes.
  • Clear validation error is raised if indexes differ.

2) Strategy trading API extended with optional symbol=

• Strategy.buy(...) and Strategy.sell(...) now accept optional symbol.
• Single-asset behavior is unchanged: symbol is optional and legacy code still works.

3) Data access expanded for per-symbol use

• self.data supports per-symbol access in multi-asset strategies:
  • self.data['AAPL'].Close[-1]
  • self.data.get('BTC-USD')
• Existing single-asset access patterns remain valid:
  • self.data.Close[-1]

4) Position / order / trade internals are symbol-aware

• Orders and trades now carry symbol metadata.
• Broker execution logic is symbol-scoped for order matching and position management.
• self.position['AAPL'] provides per-symbol position view.
• self.position.size remains portfolio aggregate (legacy semantics preserved).

5) Stats remain portfolio-level; trade table adds symbol attribution

• Existing top-level stats format remains unchanged (Return, Sharpe, Max Drawdown, etc.).
• Multi-asset runs aggregate performance at portfolio level.
• stats._trades now includes Symbol in multi-asset mode for attribution.

6) Plotting supports multi-asset runs

• bt.plot() signature remains compatible.
• Multi-asset plotting now includes:
  • Total equity curve
  • Per-asset normalized price section
  • Symbol-attributed trade markers
• Single-asset plots remain visually and behaviorally consistent.

7) Compatibility fix for FractionalBacktest

• FractionalBacktest.run() now patches both internal _data and _asset_data paths so fractional scaling continues to work correctly after multi-asset integration.

Backward compatibility

• Existing single-asset calls and strategies continue to work unchanged.
• Existing .run(), .plot(), and stats consumption patterns remain valid.
• Legacy single-asset strategy code does not need modification.

Tests

Added focused coverage for:

• Multi-asset run correctness and Symbol propagation in _trades
• Per-symbol position[...] access + aggregate portfolio position behavior
• Misaligned multi-asset index validation

Also verified key existing single-asset tests continue to pass.

Notes

• Multi-asset optimize path currently prioritizes correctness/backward safety over multiprocessing complexity in this iteration.
• Portfolio settings (cash, commission, margin, etc.) remain shared across assets