[Model] Add π0.5 (Pi0.5) VLA model support (#6950)

Signed-off-by: chenchaox <chenchaox@nvidia.com>

tests/model_tests/diffusion/test_alignment.py

@@ -82,6 +82,7 @@ EXCLUDED_MODELS = [
     "StableDiffusionXLPipeline",
     "Gr00tN1d7Pipeline",
     "Pi0Pipeline",
+    "Pi05Pipeline",
     "SanaWmPipeline",
     # Audio pipeline conditioned by a separate Qwen2.5-Omni encoder stage;
     # covered by tests/diffusion/models/auk (CPU pipeline tests, parity) and