fix: [missing-model] ollama-cloud: deepseek-v4.1-flash

[YongYuH]

Summary

• Add Ollama Cloud's currently served deepseek-v4.1-flash model
• Inherit shared DeepSeek V4.1 Flash metadata via base_model = "deepseek/deepseek-v4.1-flash"
• Record Ollama-specific reasoning controls and the API-verified context window

Evidence

• Ollama's public model API (POST https://ollama.com/api/show) lists the model as active with completion, thinking, tools, and vision capabilities
• The API reports a 1,048,576-token context window (deepseek_v41.context_length = 1048576)
• The library page (https://ollama.com/library/deepseek-v4.1-flash) confirms text and image input
• Reasoning controls follow the host surface: native think = true|false toggle plus reasoning_effort with low|high|max on the OpenAI-compatible endpoint (Flash maps requested low to low, unlike Pro)

Validation

• bun validate

[github-actions]

No actionable findings.