Re-enable `node-cli` tests in CI and fix them (#14093)

With the recent change of re-running flaky tests this should be fine!

substrate/scripts/ci/gitlab/pipeline/test.yml

@@ -216,7 +216,6 @@ test-linux-stable:
     - rusty-cachier snapshot create
     # this job runs all tests in former runtime-benchmarks, frame-staking and wasmtime tests
     # tests are partitioned by nextest and executed in parallel on $CI_NODE_TOTAL runners
-    # node-cli is excluded until https://github.com/paritytech/substrate/issues/11321 fixed
     - echo "Node index - ${CI_NODE_INDEX}. Total amount - ${CI_NODE_TOTAL}"
     - time cargo nextest run --workspace
       --locked
@@ -224,7 +223,6 @@ test-linux-stable:
       --verbose
       --features runtime-benchmarks
       --manifest-path ./bin/node/cli/Cargo.toml
-      --exclude node-cli
       --partition count:${CI_NODE_INDEX}/${CI_NODE_TOTAL}
     # we need to update cache only from one job
     - if [ ${CI_NODE_INDEX} == 1 ]; then rusty-cachier cache upload; fi

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

@@ -64,7 +64,7 @@ pub fn start_node() -> Child {
 	Command::new(cargo_bin("substrate"))
 		.stdout(process::Stdio::piped())
 		.stderr(process::Stdio::piped())
-		.args(&["--dev", "--tmp", "--ws-port=45789", "--no-hardware-benchmarks"])
+		.args(&["--dev", "--tmp", "--rpc-port=45789", "--no-hardware-benchmarks"])
 		.spawn()
 		.unwrap()
 }