feat: per-forward worker 模型 — 每条转发独立 frpc 配置+独立进程

核心改动（ring 协议不变，Task 本来就是 {Local,Remote,Link} 三元组）:
- process.WorkerKey: worker 键改为 'local~remote~port' 三元组
- renderForward: 每条 forward 渲染只含 1 个 proxy 的独立 frpc 配置
  (替代 renderRemote 把该 remote 全部 forwards 塞进一个进程)
- ClaimFn/RevokeFn: 认领/撤销只操作这一条 forward 自己的进程
- SyncWorkers/auto-start: 只拉起 localOnly 转发的独立进程,
  集群转发由 ring 认领节点拉起 (消除三台争抢 proxy already exists)
- RemoteStatus/StartRemote/StopRemote/RestartRemote: remote 级聚合辅助,
  保持 /profiles API 形状不变, 前端零改动
- handleStatus/logs: 按 worker key 解析真实 remote 名

效果: 三台节点的 frpc 各自只注册自己拥有的 proxy, 单条转发故障不再波及兄弟

internal/httpapi/server.go

@@ -257,7 +257,7 @@ func (h *Handler) handleStatus(w http.ResponseWriter, r *http.Request) {
 		binary = h.BinaryPath()
 	}
 	for _, rv := range remotes {
-		st, has := h.Process.Status(rv.Name)
+		st, has := h.Process.RemoteStatus(rv.Name)
 		fwd, _ := h.Store.LinksForRemote(rv.Name)
 		p := profileStatus{
 			Name: rv.Name, Enabled: rv.Enabled, Status: st, HasProc: has, Forwards: fwd,
@@ -271,10 +271,12 @@ func (h *Handler) handleStatus(w http.ResponseWriter, r *http.Request) {
 			}
 		}
 		// Prefer admin-derived conn state; fall back to log inference.
+		// In the per-forward model multiple workers may exist for one remote;
+		// use the first worker's log when available.
 		if p.AdminEnabled && len(p.ProxyStates) > 0 {
 			p.ConnState = connStateAdmin(p.ProxyStates)
 		} else {
-			p.ConnState = connStateOf(h.Process.LogPath(rv.Name), rv.Enabled, st)
+			p.ConnState = connStateOf(firstWorkerLog(h.Process, rv.Name), rv.Enabled, st)
 		}
 		profiles = append(profiles, p)
 	}
@@ -300,11 +302,17 @@ func (h *Handler) handleStatus(w http.ResponseWriter, r *http.Request) {
 		}
 		ts := make([]localTargetStatus, 0, len(targets))
 		for _, tg := range targets {
-			st, _ := h.Process.Status(tg.Remote)
+			// Per-forward model: this target's worker is the specific
+			// (local,remote,port) forward; active = its own process running.
+			st, has := h.Process.Status(process.WorkerKey(l.Name, tg.Remote, tg.RemotePort))
+			state := "stopped"
+			if has {
+				state = st.State
+			}
 			ts = append(ts, localTargetStatus{
 				Remote:      tg.Remote,
 				RemotePort:  tg.RemotePort,
-				WorkerState: st.State,
+				WorkerState: state,
 			})
 		}
 		localStatuses = append(localStatuses, localStatus{Local: l, Targets: ts})
@@ -346,7 +354,8 @@ func (h *Handler) handleStatus(w http.ResponseWriter, r *http.Request) {
 		if loc.LocalOnly {
 			fs.Kind = "local"
 			fs.OwnerID = selfID
-			if st, has := h.Process.Status(ln.Remote); has {
+			// Per-forward model: check this exact forward's worker process.
+			if st, has := h.Process.Status(process.WorkerKey(ln.Local, ln.Remote, ln.RemotePort)); has {
 				fs.Active = !ln.Disabled && st.State == "running"
 			}
 		} else {
@@ -727,6 +736,16 @@ func frpcVersionOf(binPath string) string {
 	return ""
 }
 
+// firstWorkerLog returns the log path of the first per-forward worker of a
+// remote (worker keys are "local~remote~port"), or "" when none exist. Used
+// by status aggregation to infer frps connectivity from worker logs.
+func firstWorkerLog(pm *process.Manager, remote string) string {
+	for _, key := range pm.RemoteWorkers(remote) {
+		return pm.LogPath(key)
+	}
+	return ""
+}
+
 // connStateOf derives the frps connection state from the local worker process
 // state plus a probe of its log. We never control the remote frps: the worker is
 // our local frpc; "connected" means frpc has successfully logged in to frps.