feat(claude): add ttl to cache_control with default 5m

Real Claude CLI traffic sends cache_control as
`{"type":"ephemeral","ttl":"1h"}`. Our previous payload only
sent `{"type":"ephemeral"}`, which is a bytewise mismatch with
the official CLI and one more third-party detection signal.

Policy: client-provided ttl is always passed through unchanged.
Proxy-generated cache_control blocks default to 5m (vs Parrot's 1h)
to avoid burning the 1h cache budget on automatic breakpoints while
still aligning with the `ttl` field being present.

  - claude/constants.go: DefaultCacheControlTTL = "5m"
  - apicompat/types.go: new AnthropicCacheControl type with TTL field;
    AnthropicTool gains optional CacheControl pointer so the mimicry
    path can attach a cache breakpoint to tools[-1] later.
  - service/gateway_service.go: anthropicCacheControlPayload gains TTL;
    marshalAnthropicSystemTextBlock and rewriteSystemForNonClaudeCode
    emit ttl=5m by default.

backend/internal/pkg/apicompat/types.go

@@ -82,10 +82,18 @@ type AnthropicImageSource struct {
 
 // AnthropicTool describes a tool available to the model.
 type AnthropicTool struct {
-	Type        string          `json:"type,omitempty"` // e.g. "web_search_20250305" for server tools
-	Name        string          `json:"name"`
-	Description string          `json:"description,omitempty"`
-	InputSchema json.RawMessage `json:"input_schema"` // JSON Schema object
+	Type         string                 `json:"type,omitempty"` // e.g. "web_search_20250305" for server tools
+	Name         string                 `json:"name"`
+	Description  string                 `json:"description,omitempty"`
+	InputSchema  json.RawMessage        `json:"input_schema"` // JSON Schema object
+	CacheControl *AnthropicCacheControl `json:"cache_control,omitempty"`
+}
+
+// AnthropicCacheControl 对应 Anthropic API 的 cache_control 字段。
+// ttl 默认由调用方决定；本项目策略见 claude.DefaultCacheControlTTL。
+type AnthropicCacheControl struct {
+	Type string `json:"type"`          // "ephemeral"
+	TTL  string `json:"ttl,omitempty"` // "5m" / "1h" / 省略=默认 5m（由 Anthropic 判定）
 }
 
 // AnthropicResponse is the non-streaming response from POST /v1/messages.

backend/internal/pkg/claude/constants.go

@@ -57,6 +57,11 @@ const APIKeyBetaHeader = BetaClaudeCode + "," + BetaInterleavedThinking + "," +
 // APIKeyHaikuBetaHeader Haiku 模型在 API-key 账号下使用的 anthropic-beta header（不包含 oauth / claude-code）
 const APIKeyHaikuBetaHeader = BetaInterleavedThinking
 
+// DefaultCacheControlTTL 是网关代理为自己生成的 cache_control 块默认使用的 ttl。
+// 真实 Claude Code CLI 当前使用 "1h"，但本仓策略是"客户端透传 ttl 优先；
+// 客户端缺省时统一使用 5m"，这样既不浪费 1h 缓存额度，也保留客户端自定义能力。
+const DefaultCacheControlTTL = "5m"
+
 // FullClaudeCodeMimicryBetas 返回最"像"真实 Claude Code CLI 的完整 beta 列表，
 // 用于 OAuth 账号伪装成 Claude Code 时使用。
 // 顺序与真实 CLI 抓包一致。