[Docs] Document DeepSeek V4 Flash Vision Exp

[zoomote]

​Opened on behalf of @taltas. Follow up by mentioning @roomote, in the web UI, or in Discord.

What changed

• Documents deepseek-v4-flash-vision-exp on the native DeepSeek provider page, including context/output limits, image billing, tools, and thinking modes.
• Documents Fireworks' provider-specific accounts/fireworks/models/deepseek-v4-flash-vision-exp entry and links its current model page.

Why this change was made

Zoo Code now supports DeepSeek's newly released experimental multimodal V4 Flash model through native DeepSeek and Fireworks. Provider documentation needs to identify the exact selectable IDs and link users to current capability and pricing details.

Authoritative sources:

• DeepSeek release announcement
• DeepSeek models and pricing
• DeepSeek vision guide
• Fireworks model page

Impact

Users can see where the model is supported, its exact IDs, its 1M-class context and 384K native output limit, and where to verify current rates. Type checks, lint, unused-import checks, and the production documentation build pass. The delegated visual-proof handoff timed out, so no screenshot evidence is included.

Related PRs

• Zoo-Code implementation: [Feat] Add DeepSeek V4 Flash Vision Exp support Zoo-Code#1488