Prefer init-provided system instruction in OpenAI Realtime

Add system_instruction parameter to the OpenAI Realtime adapter's
get_llm_invocation_params() and call _resolve_system_instruction() to
prefer init-provided over context-provided system instructions and
warn on conflicts. Previously context-provided took precedence.

src/pipecat/adapters/services/open_ai_realtime_adapter.py

@@ -43,20 +43,26 @@ class OpenAIRealtimeLLMAdapter(BaseLLMAdapter):
         """Get the identifier used in LLMSpecificMessage instances for OpenAI Realtime."""
         return "openai-realtime"
 
-    def get_llm_invocation_params(self, context: LLMContext) -> OpenAIRealtimeLLMInvocationParams:
+    def get_llm_invocation_params(
+        self, context: LLMContext, *, system_instruction: Optional[str] = None
+    ) -> OpenAIRealtimeLLMInvocationParams:
         """Get OpenAI Realtime-specific LLM invocation parameters from a universal LLM context.
 
-        This is a placeholder until support for universal LLMContext machinery is added for OpenAI Realtime.
-
         Args:
             context: The LLM context containing messages, tools, etc.
+            system_instruction: Optional system instruction from service settings.
 
         Returns:
             Dictionary of parameters for invoking OpenAI Realtime's API.
         """
         messages = self._from_universal_context_messages(self.get_messages(context))
+        effective_system = self._resolve_system_instruction(
+            messages.system_instruction,
+            system_instruction,
+            discard_context_system=True,
+        )
         return {
-            "system_instruction": messages.system_instruction,
+            "system_instruction": effective_system,
             "messages": messages.messages,
             # NOTE: LLMContext's tools are guaranteed to be a ToolsSchema (or NOT_GIVEN)
             "tools": self.from_standard_tools(context.tools) or [],

src/pipecat/services/openai/realtime/llm.py

@@ -687,14 +687,16 @@ class OpenAIRealtimeLLMService(LLMService):
         adapter: OpenAIRealtimeLLMAdapter = self.get_llm_adapter()
 
         if self._context:
-            llm_invocation_params = adapter.get_llm_invocation_params(self._context)
+            llm_invocation_params = adapter.get_llm_invocation_params(
+                self._context, system_instruction=self._settings.system_instruction
+            )
 
             # tools given in the context override the tools in the session properties
             if llm_invocation_params["tools"]:
                 settings.tools = llm_invocation_params["tools"]
 
-            # instructions in the context come from an initial "system" message in the
-            # messages list, and override instructions in the session properties
+            # The adapter resolves conflicts between init-provided and
+            # context-provided system instructions (preferring init-provided).
             if llm_invocation_params["system_instruction"]:
                 settings.instructions = llm_invocation_params["system_instruction"]