Background removal dependencies

U2-Net small (u2netp): Xuebin Qin and U-2-Net contributors. Apache License 2.0; LICENSE-U2Net.txt.
Architecture/source: https://github.com/xuebinqin/U-2-Net
ONNX conversion distributed by rembg: https://github.com/danielgatis/rembg/releases/tag/v0.0.0
Official rembg model digest: md5 8e83ca70e441ab06c318d82300c84806
Vendored model sha256: 309c8469258dda742793dce0ebea8e6dd393174f89934733ecc8b14c76f4ddd8
rembg normalization reference: Daniel Gatis and rembg contributors, MIT License; LICENSE-rembg.txt.
The ONNX model is unmodified. Our application performs browser-local preprocessing and alpha compositing.

ONNX Runtime Web 1.31.1: Microsoft Corporation, MIT License; LICENSE-ORT.txt and ThirdPartyNotices-ORT.txt.
Source: https://github.com/microsoft/onnxruntime/tree/v1.31.1/js/web
Published npm package: https://registry.npmjs.org/onnxruntime-web/-/onnxruntime-web-1.31.1.tgz
The three runtime files are unmodified. Only the WASM execution provider is used; one thread, no proxy or CDN.

The model predicts salient foreground, not certified person matting. Fine hair, glass, low contrast and multiple overlapping subjects may be imperfect. Review the PNG before use.