Fix NV-Reason-CXR quantization config error

Explicitly disable quantization parameters to prevent 'dict' object
has no attribute 'to_dict' error from embedded model config.

Added explicit parameters:
- quantization_config=None
- load_in_4bit=False
- load_in_8bit=False

Model will run at bfloat16 precision (~6-7GB VRAM).

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

medrax/tools/nv_reason_cxr.py

@@ -72,12 +72,16 @@ class NVReasonCXRTool(BaseTool):
 
             # Note: Skipping quantization due to dict.to_dict() error
             # Model will run at bfloat16 precision (~6GB memory)
+            # Explicitly disable quantization to prevent embedded config issues
             self.model = AutoModelForImageTextToText.from_pretrained(
                 model_path,
                 device_map=self.device,
                 cache_dir=cache_dir,
                 torch_dtype=torch.bfloat16,
                 trust_remote_code=True,
+                quantization_config=None,  # Explicitly disable quantization
+                load_in_4bit=False,         # Explicitly set to False
+                load_in_8bit=False,         # Explicitly set to False
             ).eval()
 
             self.processor = AutoProcessor.from_pretrained(