approval-distribution: Update topology if authorities are discovered later (#2981)

Fixes: https://github.com/paritytech/polkadot-sdk/issues/2138.

Especially on restart AuthorithyDiscovery cache is not populated so we
create an invalid topology and messages won't be routed correctly for
the entire session. This PR proposes to try to fix this by updating the
topology as soon as we now the Authority/PeerId mapping, that should
impact the situation dramatically.


[This issue was hit
yesterday](https://grafana.teleport.parity.io/goto/o9q2625Sg?orgId=1),
on Westend and resulted in stalling the finality.


# TODO

- [x] Unit tests
- [x] Test impact on versi

---------

Signed-off-by: Alexandru Gheorghe <alexandru.gheorghe@parity.io>

polkadot/node/network/statement-distribution/src/legacy_v1/mod.rs

@@ -1892,7 +1892,9 @@ pub(crate) async fn handle_network_update<Context, R>(
 				?authority_ids,
 				"Updated `AuthorityDiscoveryId`s"
 			);
-
+			topology_storage
+				.get_current_topology_mut()
+				.update_authority_ids(peer, &authority_ids);
 			// Remove the authority IDs which were previously mapped to the peer
 			// but aren't part of the new set.
 			authorities.retain(|a, p| p != &peer || authority_ids.contains(a));