Modularize qwen-format vision processors (#47573)
* draft smth
* typos evwerywhere
* fix repo
* pil backend
* fixmost tests
* and hunyuan
* fix docs and modular
* forgot to make it importable
* docs nit
* explicitly define fn, dont import
* dummy imports, find and delete
* more changes
* fix when video is padded
* fix fast CI
* Apply repo consistency fixes
* commenst
---------
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
