[Chore] Move MediaConnector to vllm.multimodal.media (#33324)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>

vllm/model_executor/models/nemotron_parse.py

@@ -58,8 +58,8 @@ from vllm.multimodal.processing import (
     PromptReplacement,
     PromptUpdate,
 )
+from vllm.tokenizers import TokenizerLike
 from vllm.transformers_utils.configs.radio import RadioConfig
-from vllm.transformers_utils.tokenizer import TokenizerLike
 from vllm.utils.tensor_schema import TensorSchema, TensorShape
 from vllm.v1.attention.backend import AttentionType