fix: suggest deepseek-v4-flash for Agent mode

[BlessedWithLuck1105]

Fixes #40851

Summary

The Agent mode model selector only suggests deepseek-v4-pro, but the official DeepSeek provider plugin also ships deepseek-v4-flash with the same agent features (tool-call, multi-tool-call, stream-tool-call per the plugin model yaml). This adds deepseek[ .-]v4[ .-]flash to agentSuggestedModelPatterns, mirroring how both gpt-5.5 and gpt-5.5-pro are suggested for OpenAI.

Screenshots

N/A (no UI change, suggestion list only)

Checklist

• This change requires a documentation update, included: No (behavioral list only, no docs page describes the suggestion list)
• I understand that this PR may be closed in case there was no previous discussion or issues. (This doesn't apply to typos!)
• I've added a test for each change that was introduced, and I tried as much as possible to make a single atomic change.
• I've updated the documentation accordingly: N/A
• I ran make lint && make type-check (backend) and vp staged (frontend) — frontend test for this file not run in this environment