docs: add NVIDIA Nemotron provider setup and reasoning variants - #1

Issue for this PR

Not required for documentation changes.

Type of change

• Bug fix
• New feature
• Refactor / code improvement
• Documentation

What does this PR do?

Add NVIDIA setup to the V2 providers page and extend V1's NVIDIA section with Nemotron 3.5 Super VL 120B A12B. Both examples link to build.nvidia.com, explicitly declare text and image input with text output, and expose seven reasoning variants, with high configured on the model as the default and none available explicitly.

How did you verify your code works?

• V1 docs production build passed; checked the generated providers page.
• V2 page rendered successfully on the running development server; generated-file check passed.
• Validated both configuration schemas and confirmed the V1 example migrates to the same model settings as V2.
• Compiled V2 requests: no variant and default send reasoning_effort: "high"; all seven variants send their corresponding values.
• No live model endpoint calls were made. Full repository checks and the V2 production build remain deferred while the V2 dev server is running.

Screenshots / recordings

Not applicable; documentation content only.

Checklist

• I have tested my changes locally
• I have not included unrelated changes in this PR