Add ZHMeanReversionStrategy ported from ZH-trading repo

[ZhuRong818]

• Port the mean reversion strategy from
  [ZH-trading/strategies/v2/meanrev.py](https://github.com/Z
  huRong818/ZH-trading/blob/main/strategies/v2/meanrev.py)
  to freqtrade IStrategy interface for backtesting
• Register the strategy in benchmark_all.py for
  inclusion in the full benchmark suite
• Strategy buys when price drops below rolling mean by
  entry_threshold, exits on reversion to mean, with
  dynamic stoploss via stop_multiple

Logic Mapping

ZH-trading	PortfolioBench (backtest)
MarketContext.mid_price	dataframe["close"]
np.mean(prices[-window:])	close.rolling(20).mean()
deviation < -entry_threshold → BUY	enter_long = 1
mid >= moving_avg - exit_threshold → EXIT	exit_long = 1
kelly_size()	freqtrade default stake (Kelly-compatible params)
stop = mid - abs(dev) * stop_multiple	custom_stoploss returns -entry_threshold × stop_multiple

Category	Timeframe	Return	Sharpe	Trades	Win Rate	Max DD
crypto	5m	+0.67%	15.22	46	65.2%	0.57%
crypto	4h	+0.16%	3.90	3	100%	0.0%
crypto	1d	+0.23%	0.05	12	58.3%	0.30%
stocks	5m	-2.57%	-45.91	48	39.6%	2.95%
stocks	4h	+0.21%	1.40	4	75.0%	0.11%
stocks	1d	+0.16%	0.04	13	53.8%	0.29%
indices	5m	-0.11%	-2.65	20	50.0%	0.42%
indices	4h	+0.27%	4.77	3	100%	0.0%
indices	1d	-0.18%	-0.04	9	44.4%	0.32%
mixed	5m	+0.55%	14.19	21	66.7%	0.23%
mixed	4h	0.00%	0	0	—	0.0%
mixed	1d	+0.31%	0.10	6	66.7%	0.11%

Key findings:

• Best performance on crypto 5m (Sharpe 15.2, 65% win
  rate) — matches the strategy's original design for
  high-frequency mean reversion on volatile assets
• Decent on longer-term stocks/indices (4h/1d) with
  positive returns but few trades
• Poor on stocks 5m — intraday stocks trend rather
  than mean-revert