MDEV-13987 Hang in FLUSH TABLES...FOR EXPORT

trx_purge_stop(): Release purge_sys->latch before attempting to
wake up the purge threads, so that they can actually wake up.
This is a regression of commit a13a636c
which attempted to fix MDEV-11802 by ensuring that srv_purge_wakeup()
will actually wait until all purge threads wake up.
Due to the purge_sys->latch, the threads might never wake up,
because some purge threads could end up waiting for purge_sys->latch
in trx_undo_prev_version_build() while holding dict_operation_lock
in shared mode. This in turn would block any DDL operations, the
InnoDB master thread, and InnoDB shutdown.

storage/innobase/srv/srv0srv.cc

@@ -2932,6 +2932,7 @@ void
 srv_purge_wakeup()
 {
 	ut_ad(!srv_read_only_mode);
+	ut_ad(!sync_check_iterate(sync_check()));
 
 	if (srv_force_recovery >= SRV_FORCE_NO_BACKGROUND) {
 		return;

storage/innobase/trx/trx0purge.cc

@@ -1781,8 +1781,8 @@ trx_purge_stop(void)
 
 		const int64_t sig_count = os_event_reset(purge_sys->event);
 		purge_sys->state = PURGE_STATE_STOP;
-		srv_purge_wakeup();
 		rw_lock_x_unlock(&purge_sys->latch);
+		srv_purge_wakeup();
 		/* Wait for purge coordinator to signal that it
 		is suspended. */
 		os_event_wait_low(purge_sys->event, sig_count);