feat: Maintain Stop Gain/Loss Prices Across LLM Decisions

[lukecold]

📝 Pull Request Template

1. Related Issue

Closes # (issue number)

2. Type of Change (select one)

Type of Change: New Feature

3. Description

This PR implements robust management for Stop Gain/Loss prices, ensuring these critical protective limits are maintained and consistently applied across trading sessions and subsequent LLM decisions.

• LLM Definition: LLM response extended to include new Stop Gain/Loss price fields.
• Persistent Storage: Defined stop prices are stored in session memory and persisted via database snapshot.
• Session Continuity: Loads persisted stop prices when resuming a session, restoring trading strategy limits.
• Informed Decisions: Existing stop prices are fed back into the LLM context to inform and assist subsequent trading actions.

4. Testing

• I have tested this locally.
• I have updated or added relevant tests.

5. Checklist

• I have read the Code of Conduct
• I have followed the Contributing Guidelines
• My changes follow the project's coding style

[vcfgv]

Thanks for the PR!

However, creating a separate table for stop_prices seems a bit heavy-handed at this stage.

Looking at the upsert_stop_price implementation, it seems to be only maintaining the current state (overwriting existing records based on strategy_id and symbol) rather than tracking any historical data.

Since the Strategy model already includes a strategy_metadata JSON column, could we simply store the stop price map there? This would significantly simplify the database schema and persistence logic.

For example, stop_prices in metadata could look like this:

{
  "stop_prices": {
    "BTC-USDT": { "stop_gain": 50000, "stop_loss": 40000, "note": "..." },
    "ETH-USDT": { "stop_gain": 3000, "stop_loss": 2000 }
  }
}

Please let me know what you think.