Refactor Object Detection Transforms

[yutong-xiang-97]

What has changed and why?

Refactored the LT-DETR object detection transform implementation into a shared ltdetr_transforms package with separate base, object detection, and utility modules. Updates all the relevant task model call sites to use the new structure while preserving the existing object detection behavior.

The new structure is split as follows:

• ltdetr_transforms/base.py contains the shared LT-DETR transform argument model, common transform pipeline construction, mosaic activation handling, and collate-function behavior that can be reused by LT-DETR-style tasks.
• ltdetr_transforms/object_detection.py contains the object-detection-specific input/output types, transform subclass, mixup/copyblend/scale-jitter handling, bbox normalization, degenerate-box filtering, and batch transform wiring.
• ltdetr_transforms/utils.py contains reusable helper functions for image-size resolution, activation schedules, albumentations transform ordering, bbox conversion/filtering, and normalization.
• The task-model transform modules now mostly provide task-specific defaults and aliases on top of the shared implementation instead of owning separate transform logic.

Also refactored the unit test accordingly.

How has it been tested?

Parity verified by AI.

Did you update CHANGELOG.md?

• Yes
• Not needed (internal change)

Did you update the documentation?

• Yes
• Not needed (internal change without effects for user)

[yutong-xiang-97]

/review