[Bot] Update Inference Providers documentation

[HuggingFaceInfra]

This PR automatically upgrades the @huggingface/tasks and @huggingface/inference packages and regenerates the Inference Providers documentation by running:

cd scripts/inference-providers
pnpm update @huggingface/tasks@latest @huggingface/inference@latest
pnpm run generate

This PR was automatically created (and merged) by the Update Inference Providers Documentation workflow.

---

Note

Low Risk
Documentation-only, script-generated markdown with no runtime or API behavior changes in this repo.

Overview
Regenerates Inference Providers docs after bumping @huggingface/tasks and @huggingface/inference, so example InferenceSnippet model IDs match current warm/trending provider mappings.

HF Inference examples now use ibm-granite/granite-embedding-311m-multilingual-r2 (feature extraction), deepset/roberta-base-squad2 (question answering), and Babelscape/wikineural-multilingual-ner (token classification) instead of the previous models.

Hyperbolic gains a Chat Completion (VLM) section using Qwen/Qwen2.5-VL-7B-Instruct, and that provider is added to the shared VLM snippet maps on chat-completion and image-text-to-text.

Together’s VLM example switches from MiniMaxAI/MiniMax-M3 to moonshotai/Kimi-K2.7-Code in both the provider page and the multi-provider VLM snippets.

^{Reviewed by Cursor Bugbot for commit 183b554. Bugbot is set up for automated code reviews on this repo. Configure here.}

[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.