[V0][Metrics] Deprecated duplicate queue time metric

vllm:time_in_queue_requests appears to be an exact duplicate
of vllm:request_queue_time_seconds.

Both record first_scheduled_time-arrival_time:

```
if seq_group.is_finished():
    time_queue_requests.append(
        seq_group.metrics.first_scheduled_time -
        seq_group.metrics.arrival_time)
```

```
def maybe_set_first_scheduled_time(self, time: float) -> None:
    if self.metrics.first_scheduled_time is None:
        self.metrics.first_scheduled_time = time
        self.metrics.time_in_queue = time - self.metrics.arrival_time
```

vllm:time_in_queue_requests was added by vllm-project#9659 and
vllm:request_queue_time_seconds was later added by vllm-project#4464. However,
neither existed when each PR was first created.

The latter seems like the right one to keep since it is implemented
in V1, used in the Grafana dashboard, and has test coverage.

Signed-off-by: Mark McLoughlin <markmc@redhat.com>

Author: markmc

vllm/engine/metrics.py

@@ -181,10 +181,13 @@ def __init__(self, labelnames: List[str], vllm_config: VllmConfig):
             "Histogram of time spent in DECODE phase for request.",
             labelnames=labelnames,
             buckets=request_latency_buckets)
+        # Deprecated in 0.8 - duplicates vllm:request_queue_time_seconds:
+        # TODO: in 0.9, only enable if show_hidden_metrics=True
         self.histogram_time_in_queue_request = self._histogram_cls(
             name="vllm:time_in_queue_requests",
-            documentation=
-            "Histogram of time the request spent in the queue in seconds.",
+            documentation=(
+                "Histogram of time the request spent in the queue in seconds. "
+                "DEPRECATED: use vllm:request_queue_time_seconds instead."),
             labelnames=labelnames,
             buckets=request_latency_buckets)
         self.histogram_model_forward_time_request = self._histogram_cls(