Implemented fast image processor for VitPose

[msminhas93]

What does this PR do?

This PR implements a fast PyTorch image processor for VitPose (VitPoseImageProcessorFast) that extends the existing BaseImageProcessorFast class. The new processor provides optimized image processing capabilities specifically designed for pose estimation tasks using the VitPose model.

Key Features

• Fast PyTorch-native implementation: Leverages PyTorch operations for efficient image processing
• Affine transformations: Supports image cropping and resizing based on bounding boxes
• Channel dimension handling: Automatic detection and conversion between different channel formats (CHW vs HWC)
• Keypoint processing: Includes methods for heatmap-based keypoint extraction and refinement
• Post-processing utilities: Dark unbiased data processing and coordinate transformation functions
• Optimized normalization: Handles variable channel counts (1, 3, 4 channels) with proper mean/std scaling

Before submitting

• This PR implements a new fast image processor for VitPose
• Did you read the contributor guideline, Pull Request section?
• Was this discussed/approved via a Github issue or the forum? Please add a link to it if that's the case. [Contributions Welcome] Add Fast Image Processors #36978
• Did you make sure to update the documentation with your changes? Here are the documentation guidelines, and here are tips on formatting docstrings.
• Did you write any new necessary tests?

Who can review?

@yonigozlan

[github-actions]

[For maintainers] Suggested jobs to run (before merge)

run-slow: auto, vitpose

[msminhas93]

Hi @yonigozlan! Could you please take a look. Thank you.