[Refactor] Decouple TimingContext from InputProcessingContext (#35083)

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

vllm/model_executor/models/paligemma.py

@@ -21,16 +21,17 @@ from vllm.multimodal.parse import (
     ImageEmbeddingItems,
     ImageProcessorItems,
     MultiModalDataItems,
-    MultiModalUUIDItems,
 )
 from vllm.multimodal.processing import (
     BaseDummyInputsBuilder,
     BaseMultiModalProcessor,
     BaseProcessingInfo,
+    ProcessorInputs,
     PromptIndexTargets,
     PromptInsertion,
     PromptUpdate,
     PromptUpdateDetails,
+    TimingContext,
 )
 from vllm.renderers import TokenizeParams
 from vllm.sequence import IntermediateTensors
@@ -228,19 +229,10 @@ class PaliGemmaMultiModalProcessor(BaseMultiModalProcessor[PaliGemmaProcessingIn
 
     def apply(
         self,
-        prompt: str | list[int],
-        mm_items: MultiModalDataItems,
-        mm_uuid_items: MultiModalUUIDItems | None = None,
-        hf_processor_mm_kwargs: Mapping[str, object] | None = None,
-        tokenization_kwargs: Mapping[str, object] | None = None,
+        inputs: ProcessorInputs,
+        timing_ctx: TimingContext,
     ) -> MultiModalInputs:
-        mm_inputs = super().apply(
-            prompt,
-            mm_items,
-            mm_uuid_items,
-            hf_processor_mm_kwargs=hf_processor_mm_kwargs,
-            tokenization_kwargs=tokenization_kwargs,
-        )
+        mm_inputs = super().apply(inputs, timing_ctx)
         prompt_token_ids = mm_inputs["prompt_token_ids"]
 
         tokenizer = self.info.get_tokenizer()