test(graphql): Remove explicit cleanup_resources

## Description
These were originally added to try and deal with a deadlock issue, which
was actually due to an issue with `inotify` on macOS, which is fixed by
#19195.

These calls are safe to remove (also note that if the test failed, they
would never be hit, because the assert would cause a panic before they
ran).

## Test plan

Rebase on top of #19195, run the following on macOS:

```
sui$ cargo nextest run       \
  -j 1 -p sui-graphql-rpc    \
  --test e2e_tests           \
  --features pg_integration
```

This used to intermittently hang, but now succeeds.

crates/sui-graphql-rpc/src/server/builder.rs

@@ -892,8 +892,6 @@ pub mod tests {
             delay.as_secs_f32()
         );
         assert_eq!(errs, vec![exp]);
-
-        cluster.cleanup_resources().await
     }
 
     #[tokio::test]
@@ -1144,7 +1142,6 @@ pub mod tests {
         let url_with_param = format!("{}?max_checkpoint_lag_ms=1", url);
         let resp = reqwest::get(&url_with_param).await.unwrap();
         assert_eq!(resp.status(), reqwest::StatusCode::GATEWAY_TIMEOUT);
-        cluster.cleanup_resources().await
     }
 
     /// Execute a GraphQL request with `limits` in place, expecting an error to be returned.

crates/sui-graphql-rpc/src/test_infra/cluster.rs

@@ -426,20 +426,6 @@ impl Cluster {
     pub async fn wait_for_checkpoint_pruned(&self, checkpoint: u64, base_timeout: Duration) {
         wait_for_graphql_checkpoint_pruned(&self.graphql_client, checkpoint, base_timeout).await
     }
-
-    /// Sends a cancellation signal to the graphql and indexer services and waits for them to
-    /// shutdown.
-    pub async fn cleanup_resources(self) {
-        self.network.cleanup_resources().await;
-        let _ = self.graphql_server_join_handle.await;
-    }
-}
-
-impl NetworkCluster {
-    pub async fn cleanup_resources(self) {
-        self.cancellation_token.cancel();
-        let _ = self.indexer_join_handle.await;
-    }
 }
 
 impl ExecutorCluster {

crates/sui-graphql-rpc/tests/e2e_tests.rs

@@ -62,7 +62,6 @@ mod tests {
             chain_id_actual
         );
         assert_eq!(&format!("{}", res), &exp);
-        cluster.cleanup_resources().await
     }
 
     #[tokio::test]
@@ -378,7 +377,6 @@ mod tests {
             .as_str()
             .unwrap();
         assert_eq!(sender_read, sender.to_string());
-        cluster.cleanup_resources().await
     }
 
     #[tokio::test]
@@ -495,7 +493,6 @@ mod tests {
         let binding = res.response_body().data.clone().into_json().unwrap();
         let res = binding.get("verifyZkloginSignature").unwrap();
         assert_eq!(res.get("success").unwrap(), false);
-        cluster.cleanup_resources().await
     }
 
     // TODO: add more test cases for transaction execution/dry run in transactional test runner.
@@ -589,7 +586,6 @@ mod tests {
             .unwrap();
         assert_eq!(sender_read, sender.to_string());
         assert!(res.get("results").unwrap().is_array());
-        cluster.cleanup_resources().await
     }
 
     // Test dry run where the transaction kind is provided instead of the full transaction.
@@ -660,7 +656,6 @@ mod tests {
         // in which case the sender is null.
         assert!(sender_read.is_null());
         assert!(res.get("results").unwrap().is_array());
-        cluster.cleanup_resources().await
     }
 
     // Test that we can handle dry run with failures at execution stage too.
@@ -750,8 +745,6 @@ mod tests {
             .as_str()
             .unwrap()
             .contains("UnusedValueWithoutDrop"));
-
-        cluster.cleanup_resources().await
     }
 
     #[tokio::test]
@@ -793,7 +786,6 @@ mod tests {
             .get("liveObjectSetDigest")
             .unwrap()
             .is_null());
-        cluster.cleanup_resources().await
     }
 
     #[tokio::test]
@@ -860,7 +852,6 @@ mod tests {
             .await
             .unwrap();
 
-        assert!(res.errors().is_empty());
-        cluster.cleanup_resources().await
+        assert!(res.errors().is_empty(), "{:#?}", res.errors());
     }
 }