[Core] Give the OpenAI Realtime wire codec a reusable home (RFC #6592 P0a) (#7640)

Signed-off-by: psv666 <2693925048@qq.com>

vllm_omni/entrypoints/duplex/realtime_input.py

@@ -5,7 +5,9 @@
 
 Everything that needed session state now lives engine-side
 (``vllm_omni.engine.duplex.realtime_commands`` for the command mapping,
-``vllm_omni.engine.duplex.realtime_events`` for the projection state). What is
+``vllm_omni.engine.duplex.realtime_events`` for the projection state), and the
+model-agnostic parsing both of those build on lives in
+``vllm_omni.protocol.realtime``. What is
 left here is the per-connection handshake policy (query-param defaults,
 autostart / resume-only rules, ``session.resume`` parsing), the wire defaults
 used to translate appends, and error rendering through the typed
@@ -20,10 +22,8 @@ from typing import TYPE_CHECKING
 
 from vllm_omni.engine.duplex.commands import DuplexCommand, DuplexCommandError
 from vllm_omni.engine.duplex.events import error_event
-from vllm_omni.engine.duplex.realtime_commands import (
-    RealtimeInputDefaults,
-    translate_realtime_command,
-)
+from vllm_omni.engine.duplex.realtime_commands import translate_realtime_command
+from vllm_omni.protocol.duplex import RealtimeInputDefaults
 
 if TYPE_CHECKING:
     from starlette.websockets import WebSocket