Fix README scenario count: 29 → 31 (18 in batch eval results) (#41)

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: antoinezambelli

README.md

@@ -141,7 +141,7 @@ python -m pytest tests/ --cov=forge --cov-report=term-missing
 
 ## Eval Harness
 
-29 scenarios measuring how reliably a model + backend combo navigates multi-step tool-calling workflows. See [Eval Guide](docs/EVAL_GUIDE.md) for full CLI reference.
+31 scenarios (18 in batch eval results) measuring how reliably a model + backend combo navigates multi-step tool-calling workflows. See [Eval Guide](docs/EVAL_GUIDE.md) for full CLI reference.
 
 ```bash
 # Ollama