Add support to export XNNPACK based static_llama (#7535)

Summary:

Add support to export XNNPACK based static_llama
- static_llama is the QNN backend hybrid/prefill+decode model with KV cache as the inference input
  - https://www.internalfb.com/code/fbsource/fbcode/executorch/examples/qualcomm/oss_scripts/llama2/model/static_llama.py

Reviewed By: tarun292

Differential Revision: D67867190

Author: Di Xu (SWE)

examples/models/llama/export_llama_lib.py

@@ -79,7 +79,14 @@
 verbosity_setting = None
 
 
-EXECUTORCH_DEFINED_MODELS = ["stories110m", "llama2", "llama3", "llama3_1", "llama3_2"]
+EXECUTORCH_DEFINED_MODELS = [
+    "stories110m", 
+    "llama2", 
+    "llama3", 
+    "llama3_1", 
+    "llama3_2", 
+    "static_llama",
+]
 TORCHTUNE_DEFINED_MODELS = ["llama3_2_vision"]