[DOCS] VLM-based image annotation pipeline (FastVLM) - new config, tool, and pipeline docs

[inference-gateway-maintainer]

Summary

Track documentation needed for the VLM-based image annotation pipeline feature (inference-gateway/cli issue #895).

Affected Pages

When the feature is implemented, the following docs pages will need updates:

1. configuration-reference.md - New vision.annotation config section (provider, model, endpoint, prompt, max_tokens, cache settings)
2. tools-reference.md - New annotate_image tool documentation
3. commands-reference.md - Any new CLI flags or subcommands for vision annotation
4. computer-use.md (if it exists) or a new vision-annotation.md page - How the annotation pipeline works, how to set up FastVLM sidecar, how to use text-only models with images

Related

• Feature request: [FEATURE] Vision annotation pipeline: let cheaper/text-only models understand screen and camera frames cli#895
• Apple FastVLM: https://github.com/apple/ml-fastvlm