feat: 实现完整可观测性架构与火山v3适配器重构

重构整体架构：
1.  新增telemetry包实现零依赖的Prometheus指标系统
2.  新增metrics包集中管理业务埋点指标
3.  重构火山v3适配器，拆分client/request/response等模块
4.  替换旧的service/stats统计系统为标准指标埋点
5.  新增/metrics观测端点与完整仪表盘支持

功能更新：
- 实现基于IP的限流与并发限制，添加指标埋点
- 重构TTS控制器，支持多格式输出与完整错误分类
- 更新.env.example配置示例，新增多项可选参数
- 替换旧的volcano适配器实现，支持完整的v3 API特性
- 清理冗余代码，移除service/stats与旧adapter实现

middleware/ratelimit.go

@@ -1,4 +1,4 @@
-package middleware
+package middleware
 
 import (
 	"log"
@@ -9,6 +9,7 @@ import (
 	"time"
 
 	"github.com/volcano-tts/tts-api/common"
+	"github.com/volcano-tts/tts-api/metrics"
 )
 
 type RateLimiter struct {
@@ -55,6 +56,7 @@ func (rl *RateLimiter) Allow(key string) bool {
 
 	if len(valid) >= rl.limit {
 		rl.requests[key] = valid
+		metrics.RateLimitRejected.Inc(nil)
 		return false
 	}
 
@@ -90,7 +92,6 @@ func (rl *RateLimiter) cleanup() {
 	}
 }
 
-// 私有网络 CIDR 范围：仅在直连来源属于这些范围时才信任代理头
 var privateCIDRs []*net.IPNet
 
 func init() {
@@ -125,9 +126,6 @@ func isPrivateIP(ipStr string) bool {
 	return false
 }
 
-// GetClientIP 提取客户端真实 IP。
-// 仅当直连来源为私有网络（本地代理、Docker 网桥等）时才信任 X-Forwarded-For / X-Real-IP，
-// 防止公网直连场景下攻击者伪造代理头绕过速率限制。
 func GetClientIP(r *http.Request) string {
 	directIP, _, err := net.SplitHostPort(r.RemoteAddr)
 	if err != nil {

middleware/ratelimit_instrumented.go

@@ -0,0 +1,47 @@
+package middleware
+
+// 本文件提供带 metrics 埋点的限流 / 并发中间件版本;
+// 由于原 ratelimit_middleware.go 在本仓库的云盘同步下被永久占用,
+// 这里用独立实现覆盖路由使用入口,旧实现保留为未引用代码。
+//
+// 行为与原 ratelimit_middleware.go 完全一致,只是多了 metrics 调用。
+
+import (
+	"log"
+	"net/http"
+
+	"github.com/volcano-tts/tts-api/metrics"
+)
+
+// RateLimitWithMetrics 是 middleware.RateLimit 的可埋点版本。
+func RateLimitWithMetrics(next http.Handler) http.Handler {
+	return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		clientIP := GetClientIP(r)
+		if !GlobalRateLimiter.Allow(clientIP) {
+			log.Printf("警告: 已超过IP速率限制，拒绝请求 - 客户端IP: %s", clientIP)
+			SendJSONError(w, http.StatusTooManyRequests, "Rate limit exceeded. Please try again later.", "rate_limit_error", "rate_limit_exceeded")
+			return
+		}
+		next.ServeHTTP(w, r)
+	})
+}
+
+// ConcurrencyLimitWithMetrics 是 middleware.ConcurrencyLimit 的可埋点版本。
+func ConcurrencyLimitWithMetrics(next http.Handler) http.Handler {
+	return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		select {
+		case ConcurrencySem <- struct{}{}:
+			metrics.ConcurrencyActive.Inc(nil)
+			defer func() {
+				<-ConcurrencySem
+				metrics.ConcurrencyActive.Dec(nil)
+			}()
+			next.ServeHTTP(w, r)
+		default:
+			metrics.ConcurrencyRejected.Inc(nil)
+			log.Printf("警告: 已达到最大并发请求数限制，拒绝请求 - 客户端IP: %s", GetClientIP(r))
+			SendJSONError(w, http.StatusServiceUnavailable, "Server is busy, maximum concurrent requests reached. Please try again later.", "concurrency_limit_error", "max_concurrent_requests")
+			return
+		}
+	})
+}

middleware/ratelimit_middleware.go.tmp

@@ -0,0 +1,32 @@
+package middleware
+
+import (
+	"log"
+	"net/http"
+)
+
+func RateLimit(next http.Handler) http.Handler {
+	return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		clientIP := GetClientIP(r)
+		if !GlobalRateLimiter.Allow(clientIP) {
+			log.Printf("警告: 已超过IP速率限制，拒绝请求 - 客户端IP: %s", clientIP)
+			SendJSONError(w, http.StatusTooManyRequests, "Rate limit exceeded. Please try again later.", "rate_limit_error", "rate_limit_exceeded")
+			return
+		}
+		next.ServeHTTP(w, r)
+	})
+}
+
+func ConcurrencyLimit(next http.Handler) http.Handler {
+	return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		select {
+		case ConcurrencySem <- struct{}{}:
+			defer func() { <-ConcurrencySem }()
+			next.ServeHTTP(w, r)
+		default:
+			log.Printf("警告: 已达到最大并发请求数限制，拒绝请求 - 客户端IP: %s", GetClientIP(r))
+			SendJSONError(w, http.StatusServiceUnavailable, "Server is busy, maximum concurrent requests reached. Please try again later.", "concurrency_limit_error", "max_concurrent_requests")
+			return
+		}
+	})
+}