test-utils: Fix wait_for_blocks in presence of reorgs (#14215)

In the cases where a reorg happens we might receive notifications
for different blocks at the same level, so instead of the chain having
count new blocks it has less and that will break the tests which use this
function.

So, use the block number to identify that `count` blocks have been built in the
chain.

Examples where this issue was hit:
  https://github.com/paritytech/polkadot/issues/7267

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

substrate/test-utils/client/src/lib.rs

@@ -42,7 +42,11 @@ use sc_client_api::BlockchainEvents;
 use sc_service::client::{ClientConfig, LocalCallExecutor};
 use serde::Deserialize;
 use sp_core::{storage::ChildInfo, testing::TaskExecutor};
-use sp_runtime::{codec::Encode, traits::Block as BlockT, OpaqueExtrinsic};
+use sp_runtime::{
+	codec::Encode,
+	traits::{Block as BlockT, Header},
+	OpaqueExtrinsic,
+};
 use std::{
 	collections::{HashMap, HashSet},
 	pin::Pin,
@@ -410,7 +414,7 @@ where
 		Box::pin(async move {
 			while let Some(notification) = import_notification_stream.next().await {
 				if notification.is_new_best {
-					blocks.insert(notification.hash);
+					blocks.insert(*notification.header.number());
 					if blocks.len() == count {
 						break
 					}