Consider making maxTokens configurable based on model capabilities

[DimitriGilbert]

CodeRabbit
Consider making maxTokens configurable based on model capabilities

The hardcoded maxTokens: 4096 might not be suitable for all models. Some models support more, while others support less.

+      // Get model-specific max tokens or use a reasonable default
+      const modelMaxTokens = modelInstance.maxOutputTokens || 4096;
+      const maxTokens = Math.min(modelMaxTokens, 4096); // Cap at 4096 for code generation
+      
       const result = await AIService.generateCompletion({
         model: modelInstance,
         system: systemPrompt,
         messages: [{ role: "user", content: userPrompt }],
         temperature: 0.1,
-        maxTokens: 4096,
+        maxTokens: maxTokens,
       });

settings... i might have to ask suno for something...