Centralize system message handling in adapters; add developer message support

Two goals:

1. Centralize system_instruction vs context system message resolution into
   the LLM adapters. This eliminates duplication between in-pipeline and
   out-of-band (run_inference) code paths across ~16 locations in service
   llm.py files.

2. Add support for "developer" role messages in conversation context, which
   is facilitated by the above centralization.

Shared helpers on BaseLLMAdapter:
- _extract_initial_system_or_developer: extracts/converts messages[0]
  based on role and whether system_instruction is provided
- _resolve_system_instruction: warns on conflicts between system_instruction
  and context system messages, returns the effective instruction

Developer message handling (new):
- Non-OpenAI adapters: an initial "developer" message is promoted to the
  system instruction when no system_instruction is provided; otherwise it
  is converted to "user". Subsequent "developer" messages are always
  converted to "user". No conflict warning is emitted for developer
  messages (unlike "system" messages).
- OpenAI adapter: "developer" messages pass through in conversation
  history without triggering conflict warnings.
- OpenAI Responses adapter: "developer" messages are kept as "developer"
  role (same as "system", which is also converted to "developer" for the
  Responses API).

Other behavior changes:
- Gemini: "initial" system message detection now checks messages[0] only
  (previously searched anywhere in the list)
- Bedrock: a lone system message is now converted to "user" instead of
  being extracted to an empty message list (matches existing Anthropic
  behavior)

tests/test_get_llm_invocation_params.py

@@ -424,10 +424,11 @@ class TestGeminiGetLLMInvocationParams(unittest.TestCase):
         context = LLMContext(messages=messages)
         params = self.adapter.get_llm_invocation_params(context)
 
-        # System instruction should be extracted
-        self.assertEqual(params["system_instruction"], "You are a helpful assistant.")
+        # When there's only one message, it's converted to user in-place (not extracted)
+        # so system_instruction is None
+        self.assertIsNone(params["system_instruction"])
 
-        # But since there are no other messages, it should also be added back as a user message
+        # The system message should be converted to a user message
         self.assertEqual(len(params["messages"]), 1)
         self.assertEqual(params["messages"][0].role, "user")
         self.assertEqual(params["messages"][0].parts[0].text, "You are a helpful assistant.")
@@ -973,7 +974,7 @@ class TestAWSBedrockGetLLMInvocationParams(unittest.TestCase):
         self.assertEqual(params["messages"][2]["content"][0]["text"], "Remember to be concise.")
 
     def test_single_system_message_handling(self):
-        """Test that a single system message is extracted as system parameter and no messages remain."""
+        """Test that a single system message is converted to user role when no other messages exist."""
         messages = [
             {"role": "system", "content": "You are a helpful assistant."},
         ]
@@ -984,13 +985,16 @@ class TestAWSBedrockGetLLMInvocationParams(unittest.TestCase):
         # Get invocation params
         params = self.adapter.get_llm_invocation_params(context)
 
-        # System should be extracted (in AWS Bedrock format)
-        self.assertIsInstance(params["system"], list)
-        self.assertEqual(len(params["system"]), 1)
-        self.assertEqual(params["system"][0]["text"], "You are a helpful assistant.")
+        # When there's only one message, it's converted to user in-place (not extracted)
+        # so system is None
+        self.assertIsNone(params["system"])
 
-        # No messages should remain after system extraction
-        self.assertEqual(len(params["messages"]), 0)
+        # Single system message should be converted to user role
+        self.assertEqual(len(params["messages"]), 1)
+        self.assertEqual(params["messages"][0]["role"], "user")
+        self.assertEqual(
+            params["messages"][0]["content"][0]["text"], "You are a helpful assistant."
+        )
 
 
 class TestPerplexityGetLLMInvocationParams(unittest.TestCase):