Add WaitGroup.WaitUnlock(l Locker),

The passed in `Locker` will be called immediately if no need to wait, or right before wait(`runtime_Semacquire`).

This can be useful when some resources can be released before wait, e.g, when `Add/Wait` are synchronised by RWMutex,
where `Add` only happens while holding `RLock` and `Wait` only happens while holding `Lock`.

Here's a concrete example:

https://github.com/zhiqiangxu/util/blob/master/closer/state.go#L58

Fixed #36349

Author: blockchaindevsh

src/sync/waitgroup.go

@@ -101,6 +101,11 @@ func (wg *WaitGroup) Done() {
 
 // Wait blocks until the WaitGroup counter is zero.
 func (wg *WaitGroup) Wait() {
+	wg.WaitUnlock(nil)
+}
+
+// if Locker is not nil, it will be Unlock immediately if no need to wait, or right before wait.
+func (wg *WaitGroup) WaitUnlock(l Locker) {
 	statep, semap := wg.state()
 	if race.Enabled {
 		_ = *statep // trigger nil deref early
@@ -116,6 +121,9 @@ func (wg *WaitGroup) Wait() {
 				race.Enable()
 				race.Acquire(unsafe.Pointer(wg))
 			}
+			if l != nil {
+				l.Unlock()
+			}
 			return
 		}
 		// Increment waiters count.
@@ -127,6 +135,9 @@ func (wg *WaitGroup) Wait() {
 				// otherwise concurrent Waits will race with each other.
 				race.Write(unsafe.Pointer(semap))
 			}
+			if l != nil {
+				l.Unlock()
+			}
 			runtime_Semacquire(semap)
 			if *statep != 0 {
 				panic("sync: WaitGroup is reused before previous Wait has returned")