Dynamic and Batch Support for Object Detector

[kevchengcodes]

Summary

Enables the ObjectDetector to run dynamic models as well as static models with batch_size > 1.

Changes include resolving dynamic dimensions similar to how llm-runner does it, constructing a batched input tensor based on the list of input images, and post-processing each output.

Models can have dynamic batch size, image input height/width. Note that batched images will be resized to a common size when constructing the (B,C,H,W) input tensor. Users can optionally input a custom input height/width for their dynamic batch.

Verification

Tested with YoloS:

uv run models/yolo/export.py --dynamic
swift run -c release object-detector --model exports/yolos-base_float32_dynamic.aimodel --image img.png --output-image out.png

New unit tests to validate batch support is working.

[carinapeng]

Thank you!