[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 implementation changes in this repo.

Overview
Regenerated Inference Providers docs after bumping @huggingface/tasks and @huggingface/inference. Updates are limited to InferenceSnippet providersMapping values and which provider/task sections appear.

Example model swaps: Cohere LLM docs now use CohereLabs/tiny-aya-earth; HF Inference chat/text examples use katanemo/Arch-Router-1.5B instead of HuggingFaceBio/Carbon-3B; token-classification examples use OpenMed/OpenMed-PII-SuperClinical-Large-434M-v1 instead of dslim/bert-base-NER. Fal’s image-to-image snippet points at a Qwen image-edit LoRA route rather than FLUX.2-dev.

Provider coverage: Groq and Nscale no longer document Chat Completion (VLM) on their provider pages, and groq / nscale are dropped from aggregated VLM snippets in chat-completion and image-text-to-text task docs. LLM and other task mappings are aligned across provider and task pages.

^{Reviewed by Cursor Bugbot for commit 8293b58. Bugbot is set up for automated code reviews on this repo. Configure here.}