fix: separate Anthropic output and context limits

Direct Anthropic connections reused the configured context window as max_tokens, which could impose an unintended output value or send a provider-invalid request. Anthropic connections now expose an independent per-response maximum output setting, defaulting to 32K for the protocol-required field, while OpenAI direct requests remain without an added output token cap and direct tool rounds remain unmetered.

Release note: Anthropic 直连模型现在可单独配置单次最大输出，不再复用上下文窗口；OpenAI 直连请求和工具调用轮次不增加额外配额。

src/shared/contracts.ts

@@ -568,6 +568,9 @@ export {
   minimumModelContextWindowTokens
 } from './context-window'
 
+export const defaultAnthropicMaximumOutputTokens = 32_000
+export const maximumModelOutputTokens = 10_000_000
+
 const modelProfileInputSchema = z
   .object({
     id: modelProfileIdSchema,
@@ -588,6 +591,12 @@ const modelProfileInputSchema = z
       .min(minimumModelContextWindowTokens)
       .max(maximumModelContextWindowTokens)
       .optional(),
+    maximumOutputTokens: z
+      .number()
+      .int()
+      .min(1)
+      .max(maximumModelOutputTokens)
+      .optional(),
     imageGenerationQuality: imageGenerationQualitySchema,
     apiKey: modelApiKeyUpdateSchema
   })
@@ -968,6 +977,7 @@ export type ModelConnectionSettings = {
   authentication: ModelAuthentication
   supportsImageInput?: boolean
   contextWindowTokens?: number
+  maximumOutputTokens?: number
   imageGenerationQuality: ImageGenerationQuality
   apiKeyConfigured: boolean
   credentialSource: 'none' | 'encrypted' | 'environment' | 'unreadable'