Switch LLM provider to DeepSeek v4 Pro

- Create shared openai client (src/lib/openai.ts) pointing to api.deepseek.com
- Replace all gpt-4o-mini model references with deepseek-v4-pro
- Refactor 3 API routes to use shared client instead of inline OpenAI() calls
- Add OPENAI_BASE_URL env var for endpoint configuration

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

src/lib/engines/summary-generator.ts

@@ -106,7 +106,7 @@ export async function generateWeeklySummary(
 
   try {
     const response = await openai.chat.completions.create({
-      model: "gpt-4o-mini",
+      model: "deepseek-v4-pro",
       messages: [
         { role: "system", content: SUMMARY_PROMPT },
         {