fix: factor DeepSeek V4 Flash 0731 providers

[rekram1-node]

Summary

• add canonical lab metadata for the distinct DeepSeek V4 Flash 0731 release
• factor Baseten, Chutes, DeepSeek, Fireworks AI, OpenCode Zen, OpenCode Go, OpenRouter, Venice, and Vercel entries through base_model
• update both paid and free OpenCode Zen routes plus the OpenCode Go route
• update the EmpirioLabs canonical mapping to prevent sync regressions

Sources

• Model card and release description: https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0731
• DeepSeek API docs confirm the unchanged deepseek-v4-flash API ID now serves DeepSeek-V4-Flash-0731: https://api-docs.deepseek.com/quick_start/pricing/
• OpenCode Go lists the undated deepseek-v4-flash route and current DeepSeek pricing: https://opencode.ai/docs/go/
• Model card documents low, high, and max reasoning effort levels and recommends up to 384K output for high/max

Validation

• bun validate
• git diff --check

[github-actions]

No actionable findings.

[github-actions]

Action items

• [medium] [possible mistake] providers/opencode/models/deepseek-v4-flash.toml:1 - Check: Provider display name should stay a real delta when retargeting base_model. Why: This file has no name override, so switching base_model from deepseek/deepseek-v4-flash to deepseek/deepseek-v4-flash-0731 changes the resolved display name from "DeepSeek V4 Flash" to "DeepSeek V4 Flash 0731" for model id deepseek-v4-flash. That diverges from the DeepSeek first-party entry (keeps name = "DeepSeek V4 Flash") and from nearby OpenCode aliases (DeepSeek V4 Flash (New) / DeepSeek V4 Flash Free (New)). Action: Add an explicit name = "DeepSeek V4 Flash" (or another intentional product name) if the OpenCode id should keep the stable Flash label; only omit name if showing "0731" is deliberate.