Currently, we log userOpsKilled counter only when RSTL acquisition is successful and step down is success. Instead, we should log the counter after every time we stopped killing user operations.
In addition, we should record operations running and killed during step up and rollback in a manner similar to step down.
add repl.stepUp.userOperationsRunning/Killed metrics
Investigate changes in SERVER-42825: Log metrics.repl.stepDown counters after stopped killing user operations as part of state transition (step up/ step down).