Add fine-tuned SuperAnimal-Quadruped support and improve demo setup

[xiu-cs]

Summary

This PR adds first-class support for the fine-tuned SuperAnimal-Quadruped 2D pose model used by the animal pipeline, enabling direct 26-joint Animal3D keypoint prediction and automatic checkpoint download from Hugging Face. It also improves the out-of-the-box demo/install path, fixes CPU fallback for the human HRNet demo, and removes unused legacy code/assets.

Changes

• Add support for a fine-tuned SuperAnimal-Quadruped 2D checkpoint that predicts the 26-joint Animal3D layout directly.
• Auto-download animal demo checkpoints from Hugging Face on first run:
  • sa_finetune_hrnet_w32.pt for 2D animal pose
  • fmpose3d_animals.pth for the 3D lifter
• Refactor animals/demo/vis_animals.py to build the 2D estimator and 3D lifter once, then reuse them across images.
• Add SuperAnimalConfig options for fine-tuned checkpoints, detector overrides, and lazy Hugging Face resolution.
• Update animal defaults and docs from older Rat7M/legacy assumptions toward Animal3D.
• Fix human HRNet loading on CPU-only environments by using device-aware map_location and moving inputs to the model device.
• Pin install dependencies to torch>=2.4.1,<2.5 and torchvision>=0.19.1,<0.20, and document the PyTorch/CUDA behavior in the README.
• Restrict package Python metadata to >=3.10,<3.13; README recommends Python 3.10 because install/demo paths were tested there.
• Remove unused legacy animal modules and unused YOLO/HRNet assets.
• Add/update tests for the fine-tuned SuperAnimal path and config behavior.
• Add mot to the codespell ignore list.

Validation

Ran install, test, and demo checks locally:

python3 -m pip install -e '.[animals,viz]' --dry-run
python3 -m pytest tests/test_demo_human.py tests/fmpose3d_api/test_fmpose3d.py -q
python3 -m pytest tests/test_model.py tests/test_training_pipeline.py -q
bash demo/vis_in_the_wild.sh
bash animals/demo/vis_animals.sh

Results

• Human demo passes on both CPU-only and GPU paths.
• Animal demo passes and auto-resolves both Hugging Face checkpoints.
• Relevant tests pass: 78 passed for human demo/API tests, 8 passed for model/training smoke tests.