Add support for metric3d (v1 & v2)

[xenova]

Closes #1247

Example usage w/ pipeline API:

import { pipeline } from '@huggingface/transformers';

const model = 'onnx-community/metric3d-vit-small';
const depth_estimator = await pipeline('depth-estimation', model);

// Predict depth of an image
const url = 'https://huggingface.co/datasets/Xenova/transformers.js-docs/resolve/main/cats.jpg';
const { depth, predicted_depth } = await depth_estimator(url);

// Visualize the output
depth.save('depth.png');

// Or access the raw values with `predicted_depth`
console.log(predicted_depth);
// Tensor {
//   data: Float32Array(307200) [
//     1.9032787084579468, 1.8878453969955444, 1.8748314380645752,
//     ...
//   ],
//   type: 'float32',
//   dims: [ 480, 640 ],
//   size: 307200
// }

Example usage w/ AutoModel API:

import { AutoProcessor, AutoModelForDepthEstimation, RawImage } from '@huggingface/transformers';

const model_id = 'onnx-community/metric3d-vit-small';
const model = await AutoModelForDepthEstimation.from_pretrained(model_id);
const processor = await AutoProcessor.from_pretrained(model_id);

const url = 'https://huggingface.co/datasets/Xenova/transformers.js-docs/resolve/main/cats.jpg';
const image = await RawImage.fromURL(url);
const inputs = await processor(image);
const { predicted_depth, predicted_normal, normal_confidence } = await model(inputs);

console.log(predicted_depth);
console.log(predicted_normal);
console.log(normal_confidence);

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.