Avoid writing ChannelManager when hitting lnd bug 6039

When we hit lnd bug 6039, we end up sending error messages to peers
in a loop. This should be fine, but because we used the generic
`PersistenceNotifierGuard::notify_on_drop` lock above the specific
handling, we end up writing `ChannelManager` every time we manage a
round-trip to our peer.

This can add up quite quickly, and isn't actually changing, so we
really need to avoid writing the `ChannelManager` in this case.

lightning/src/ln/channelmanager.rs

@@ -9242,8 +9242,6 @@ where
 	}
 
 	fn handle_error(&self, counterparty_node_id: &PublicKey, msg: &msgs::ErrorMessage) {
-		let _persistence_guard = PersistenceNotifierGuard::notify_on_drop(self);
-
 		match &msg.data as &str {
 			"cannot co-op close channel w/ active htlcs"|
 			"link failed to shutdown" =>
@@ -9277,13 +9275,21 @@ where
 								log_level: Level::Trace,
 							}
 						});
+						// This can happen in a fairly tight loop, so we absolutely cannot trigger
+						// a `ChannelManager` write here.
+						PersistenceNotifierGuard::optionally_notify(
+							self,
+							|| -> NotifyOption { NotifyOption::SkipPersistHandleEvents }
+						);
 					}
 				}
 				return;
 			}
 			_ => {}
 		}
 
+		let _persistence_guard = PersistenceNotifierGuard::notify_on_drop(self);
+
 		if msg.channel_id.is_zero() {
 			let channel_ids: Vec<ChannelId> = {
 				let per_peer_state = self.per_peer_state.read().unwrap();