Add --llm_image_context for context-aware image descriptions

[vertgo]

New opt-in processor (LLMImageContextDescriptionProcessor) that describes every Picture/Figure block with an LLM, supplying the page's extracted markdown and the page image as context. Returns a short caption (rendered as image alt text) and a long description (rendered as a hidden HTML comment), so a text-only LLM reading the markdown alone can understand the visual content without a multimodal pipeline.

• New flag --llm_image_context, orthogonal to --use_llm; works whether or not image extraction is enabled.
• Picture/Figure blocks gain short_caption + long_description fields.
• HTML renderer emits alt text + a hidden ; markdown renderer converts that aside into an HTML comment.
• Gemini service activates for this flag independently of --use_llm.

Developed with assistance from Claude (Claude Code).

[github-actions]

CLA Assistant Lite bot:
Thank you for your submission, we really appreciate it. Like many open-source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution. You can sign the CLA by just posting a Pull Request Comment same as the below format.

---

I have read the CLA Document and I hereby sign the CLA

---

_{You can retrigger this bot by commenting recheck in this Pull Request}