Fix env.reset() to return (obs, info) tuple for SB3 v2.0+ compatibility

[he-yufeng]

Summary

Four environments' reset() methods return only the observation instead of the (obs, info) tuple required by the Gymnasium API (and SB3 v2.0+), causing ValueError: too many values to unpack (expected 2) when used with DummyVecEnv.

Fixed Environments

• env_btc_ccxt.py (BitcoinEnv)
• env_multiple_crypto.py (CryptoEnv)
• env_stocktrading_stoploss.py (StockTradingEnvStopLoss)
• env_stocktrading_cashpenalty.py (StockTradingEnvCashpenalty)

Note: env_stocktrading.py and env_stocktrading_np.py already return (obs, {}) correctly.

Change

Each fix is a single-line change: return state → return state, {}

Test plan

• Verify DummyVecEnv wrapping works with each fixed environment
• Verify training with SB3 PPO/A2C works without the unpacking error

Fixes #1051

🤖 Generated with Claude Code

[atharvajoshi01]

SB3 v2.0 changed the reset() API to return (obs, info) tuple. This fixes all the env classes to match. Clean and necessary for anyone using stable-baselines3 >= 2.0.

[he-yufeng]

The remaining pre-commit.ci - pr failure is not from the five files changed in this PR. pre-commit.ci is trying to reorder imports / run black on unrelated existing files such as processor_yahoofinance.py, env_portfolio_optimization.py, and the portfolio/elegantrl model files.

I am leaving this branch scoped to the SB3 reset API compatibility fix instead of widening it into a repository-wide formatting update.

[he-yufeng]

Any chance someone could take a look at this? It's been sitting here green for a few months — happy to rebase or adjust if anything's needed.