Merge branch 'main' into cc_perf_fixes

Author: dhansen-nvidia

tensorrt_llm/_torch/pyexecutor/llm_request.py

@@ -764,7 +764,8 @@ def executor_request_to_llm_request(
         cache_salt_id=executor_request.cache_salt_id,
         arrival_time=getattr(executor_request, "py_arrival_time", None),
         py_multimodal_data=getattr(executor_request, "py_multimodal_data",
-                                   None))
+                                   None),
+        kv_cache_retention_config=executor_request.kv_cache_retention_config)
     if child_req_ids:
         for child_id in child_req_ids:
             llm_request.create_child_request(child_id)