Convert `NVFP4BlockScaling` to dataclass (#2227)

Fix passing args to nvfp4 recipe
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

transformer_engine/common/recipe/__init__.py

@@ -382,6 +382,7 @@ class Float8BlockScaling(Recipe):
         )
 
 
+@dataclass()
 class NVFP4BlockScaling(Recipe):
     """
     Use the NVFP4 scaling strategy.