add support for 2D video pooling

update the processor to the working version for videomme

change content mask type to bool

add image and video chunk to each token chunk in the processor

fix chunking error

add prefill chunking

remove useless logs

update video processor params to the qwen3vl paper version

change default fps to 1

ignore type errors in the processing codes

adapt to transformers==4.57.6

adapt to transformers==4.57.6

initial commit