add llama sdpa to generation script

Author: lucylq

examples/models/llama2/runner/generation.py

@@ -19,8 +19,11 @@
     Message,
     Tokenizer,
 )
+
+from executorch.extension.pybindings import portable_lib
 from executorch.extension.pybindings.portable_lib import _load_for_executorch
 
+from executorch.extension.llm.custom_ops import sdpa_with_kv_cache
 
 class CompletionPrediction(TypedDict, total=False):
     generation: str