Resolves a race condition in the reconfig test. When we set a failpoint we don't necessarily know that we've hit the failpoint. On this test we set failpoints to ensure that the primary pauses its write loop, but previously did not wait to hit the failpoint before continuing the secondary's step-up attempt. As a result, it was possible for one last write to sneak through while we un-paused the secondary's step-up. If that write happened to hit the log service after the secondary's catch-up, then we'd detect the new primary prematurely and step down. This change adds a wait() to ensure that we've hit the last writes from the primary before we allow the secondary to complete its step-up