fix(client): stop holding the inbound-lock registry mutex while waiting on one inbound

lockInbound acquired the global registry mutex and then blocked on the
per-inbound mutex without releasing the registry first. A slow client
operation holding one inbound's mutex (for example a bulk delete pushing to
an unreachable node) made the next waiter park on that inbound while still
holding the registry mutex, which in turn blocked lockInbound for every
other inbound — freezing client mutations panel-wide. Release the registry
mutex before taking the per-inbound lock.

internal/web/service/client_locks.go

@@ -28,12 +28,12 @@ var (
 
 func lockInbound(inboundId int) *sync.Mutex {
 	inboundMutationLocksMu.Lock()
-	defer inboundMutationLocksMu.Unlock()
 	m, ok := inboundMutationLocks[inboundId]
 	if !ok {
 		m = &sync.Mutex{}
 		inboundMutationLocks[inboundId] = m
 	}
+	inboundMutationLocksMu.Unlock()
 	m.Lock()
 	return m
 }