Adding superglue fast image processing (#41394)
* Default implementation - no time improvement
* Improved implementation - apparently 2 times faster with only simple function refactor
* elementary torch first approach, still need further implementation of torch first method
* torch-first approach finished
* refactor processor
* refactor test
* partial doc update
* EfficientLoFTRImageProcessorFast based implementation
* EfficientLoFTRImageProcessorFast based implementation
* Logic checked - Test Passed - Validated execution speed
* use modular for efficientloftr
* fix import
---------
Co-authored-by: yonigozlan <yoni.gozlan@huggingface.co>
Co-authored-by: Yoni Gozlan <74535834+yonigozlan@users.noreply.github.com>
