[VLM] Merged multi-modal processor for InternVL-based models (#12553)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Signed-off-by: Isotr0py <2037008807@qq.com>
Co-authored-by: Isotr0py <2037008807@qq.com>

tests/models/decoder_only/vision_language/test_models.py

@@ -250,6 +250,7 @@ VLM_TEST_SETTINGS = {
         max_model_len=8192,
         dtype="bfloat16",
         use_tokenizer_eos=True,
+        num_logprobs=10,
         patch_hf_runner=model_utils.h2ovl_patch_hf_runner,
     ),
     "idefics3": VLMTestInfo(
@@ -282,7 +283,6 @@ VLM_TEST_SETTINGS = {
         dtype="bfloat16",
         use_tokenizer_eos=True,
         patch_hf_runner=model_utils.internvl_patch_hf_runner,
-        marks=[large_gpu_mark(min_gb=32)],
     ),
     "llava_next": VLMTestInfo(
         models=["llava-hf/llava-v1.6-mistral-7b-hf"],