💥 feat(agent)!: 添加主机监控代理首版

* 实现周期采集、快照缓存、本地 HTTP API 和主动上报

* 补齐 Linux 和 Windows 主机、网络、资源、进程和 TCP 采集

* 添加服务生命周期管理、配置加载、构建脚本和测试覆盖

BREAKING CHANGE: network.ips 拆分为 network.ipv4s 和 network.ipv6s。

internal/reporter/reporter.go

@@ -0,0 +1,99 @@
+package reporter
+
+import (
+	"bytes"
+	"context"
+	"encoding/json"
+	"fmt"
+	"log/slog"
+	"net/http"
+	"time"
+
+	"mfd-monitor-agent/internal/status"
+)
+
+const defaultRetryDelay = 2 * time.Second
+
+// Provider supplies the cached HostStatus used for outbound reports.
+type Provider interface {
+	// Get returns the latest snapshot and whether a snapshot is ready.
+	Get() (status.HostStatus, bool)
+}
+
+// Client posts HostStatus snapshots to a configured center endpoint.
+type Client struct {
+	endpoint string
+	client   *http.Client
+}
+
+// NewClient creates a reporting client with the provided endpoint and request timeout.
+func NewClient(endpoint string, timeout time.Duration) *Client {
+	return &Client{
+		endpoint: endpoint,
+		client:   &http.Client{Timeout: timeout},
+	}
+}
+
+// Report posts one HostStatus snapshot as JSON and returns non-2xx responses as errors.
+func (c *Client) Report(snapshot status.HostStatus) error {
+	body, err := json.Marshal(snapshot)
+	if err != nil {
+		return err
+	}
+	req, err := http.NewRequest(http.MethodPost, c.endpoint, bytes.NewReader(body))
+	if err != nil {
+		return err
+	}
+	req.Header.Set("Content-Type", "application/json")
+	resp, err := c.client.Do(req)
+	if err != nil {
+		return err
+	}
+	defer resp.Body.Close()
+	if resp.StatusCode < 200 || resp.StatusCode >= 300 {
+		return fmt.Errorf("report returned HTTP %d", resp.StatusCode)
+	}
+	return nil
+}
+
+// Run periodically posts the latest cached snapshot until ctx is canceled.
+func Run(ctx context.Context, provider Provider, client *Client, interval time.Duration, logger *slog.Logger) {
+	runWithRetryDelay(ctx, provider, client, interval, defaultRetryDelay, logger)
+}
+
+// runWithRetryDelay runs the reporting loop and retries one failed report after retryDelay.
+func runWithRetryDelay(ctx context.Context, provider Provider, client *Client, interval time.Duration, retryDelay time.Duration, logger *slog.Logger) {
+	ticker := time.NewTicker(interval)
+	defer ticker.Stop()
+	for {
+		select {
+		case <-ctx.Done():
+			return
+		case <-ticker.C:
+			snapshot, ok := provider.Get()
+			if !ok {
+				continue
+			}
+			if err := reportWithRetry(ctx, client, snapshot, retryDelay); err != nil {
+				logger.Warn("status report failed", "error", err)
+			}
+		}
+	}
+}
+
+// reportWithRetry posts a snapshot and performs one cancellation-aware retry on failure.
+func reportWithRetry(ctx context.Context, client *Client, snapshot status.HostStatus, retryDelay time.Duration) error {
+	err := client.Report(snapshot)
+	if err == nil {
+		return nil
+	}
+	select {
+	case <-ctx.Done():
+		return ctx.Err()
+	case <-time.After(retryDelay):
+	}
+	if retryErr := client.Report(snapshot); retryErr != nil {
+		return retryErr
+	}
+	return nil
+}