docs: update NVIDIA integration page

[chenopis]

Content Updates

• Removed redundant TOC — The toc: true in frontmatter already handles this
• Rewrote Overview — Clearer description of the package, API catalog, and NIM
• Added Prerequisites section — Step-by-step API key setup instructions

Model Updates

• Embedding model: nvolveqa_40k → nvidia/llama-3.2-nv-embedqa-1b-v2 (current model)
• Generator model: nv_llama2_rlhf_70b → meta/llama-3.1-70b-instruct (current model)

Typo Fixes

• floud → flour
• Omlette → Omelet

Formatting Improvements

• Consistent heading structure (## for main sections, ### for components)
• Better code block formatting with proper spacing
• Cleaner component descriptions using bold labels
• Improved code example output comments

New Section

• Added "Self-host with NVIDIA NIM" section

Code Example Cleanup

• Removed unnecessary parameters in NvidiaGenerator example (seed, bad, stop → cleaner defaults)
• Fixed data dict formatting in pipeline examples for readability
• Added missing import in RAG pipeline example (from haystack import Pipeline)

[bilgeyucel]

Thanks @chenopis for the changes! Left some minor comments 🙌

[chenopis]

Thanks @chenopis for the changes! Left some minor comments 🙌

@bilgeyucel Thanks for your feedback. I integrated your suggestions, kept the ToC and updated it. PTAL