With `gc+phases+ref=debug`, the following log is generated using Parallel GC. Note that the number of non-strong refs is <1000 (default value of `ReferencesPerThread`), so only 1 worker should be used. However, it prints "Workers: 10" (10 is the #Parallel Workers).

```
[3.284s][debug][gc,phases,ref] GC(0) Reconsider SoftReferences: 0.0ms
[3.284s][debug][gc,phases,ref] GC(0) SoftRef (ms): skipped
[3.284s][debug][gc,phases,ref] GC(0) Notify Soft/WeakReferences: 0.1ms
[3.284s][debug][gc,phases,ref] GC(0) SoftRef (ms): Min: 0.0, Avg: 0.0, Max: 0.0, Diff: 0.0, Sum: 0.0, Workers: 10
[3.284s][debug][gc,phases,ref] GC(0) WeakRef (ms): Min: 0.0, Avg: 0.0, Max: 0.1, Diff: 0.1, Sum: 0.1, Workers: 10
[3.284s][debug][gc,phases,ref] GC(0) FinalRef (ms): Min: 0.0, Avg: 0.0, Max: 0.0, Diff: 0.0, Sum: 0.0, Workers: 10
[3.284s][debug][gc,phases,ref] GC(0) Total (ms): Min: 0.0, Avg: 0.0, Max: 0.1, Diff: 0.1, Sum: 0.1, Workers: 10
[3.284s][debug][gc,phases,ref] GC(0) Notify and keep alive finalizable: 0.0ms
[3.284s][debug][gc,phases,ref] GC(0) FinalRef (ms): skipped
[3.284s][debug][gc,phases,ref] GC(0) Notify PhantomReferences: 0.0ms
[3.284s][debug][gc,phases,ref] GC(0) PhantomRef (ms): Min: 0.0, Avg: 0.0, Max: 0.0, Diff: 0.0, Sum: 0.0, Workers: 10
[3.284s][debug][gc,phases,ref] GC(0) SoftReference:
[3.284s][debug][gc,phases,ref] GC(0) Discovered: 0
[3.284s][debug][gc,phases,ref] GC(0) Cleared: 0
[3.284s][debug][gc,phases,ref] GC(0) WeakReference:
[3.284s][debug][gc,phases,ref] GC(0) Discovered: 348
[3.284s][debug][gc,phases,ref] GC(0) Cleared: 220
[3.284s][debug][gc,phases,ref] GC(0) FinalReference:
[3.284s][debug][gc,phases,ref] GC(0) Discovered: 3
[3.284s][debug][gc,phases,ref] GC(0) Cleared: 3
[3.284s][debug][gc,phases,ref] GC(0) PhantomReference:
[3.284s][debug][gc,phases,ref] GC(0) Discovered: 157
[3.284s][debug][gc,phases,ref] GC(0) Cleared: 68
```

Those lines are printed by `WorkerDataArray<T>::print_summary_on`, which has built-in logic for skipping "inactive" workers. However, when a single worker is used for ref processing, the "worker id", used by `WorkerDataArray` to decide which worker is active, is incorrect.

In `ReferenceProcessor::run_task`, the else branch is for ref processing using a single thread (sequential case):

```
  if (processing_is_mt()) {
    ...
  } else {
    for (unsigned i = 0; i < _max_num_queues; ++i) {
      proxy_task.work(i); <--- here we are passing [0.._max_num_queues-1] as "worker id"
    }
  }
```

This confuses WorkerDataArray to believe many workers are active in this subphase of ref-processing.