[Model] Add support for moonshotai/Kimi-Audio-7B-Instruct (#36127)

Signed-off-by: tunglinwood <tunglinwood@gmail.com>
Signed-off-by: tunglinwood <tomwu.tunglin@gmail.com>
Signed-off-by: tunglinwood <113751333+tunglinwood@users.noreply.github.com>

tests/models/test_initialization.py

@@ -103,6 +103,12 @@ def can_initialize(
             "pickle error when loading `transformers.models.auto.CONFIG_MAPPING`"
         )
 
+    if model_arch == "MoonshotKimiaForCausalLM":
+        pytest.skip(
+            "Kimi-Audio requires SpeechToTextConfig "
+            "which is not configured in test environment"
+        )
+
     if model_arch in ["DeepseekV32ForCausalLM", "GlmMoeDsaForCausalLM"]:
         from vllm.platforms import current_platform