test(metrics): the rollup test checks its own bucket, not the table's count (#5752)

#5752 "Container CPU and memory history lives in the one metric store".
Docker persist now records each snapshot at its own time, so test_docker's
June and early-October snapshots leave raw plugin_metrics rows older than
the 7-day window, and the rollup's table-wide counts include them (CI run
9119: 11 buckets, not 1). The test's own host is still checked exactly:
one hourly row averaging 15 over 2 samples, and one raw sample left.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

tests/integration/test_host_metrics.py

@@ -120,8 +120,11 @@ def test_rollup_aggregates_old_raw_and_prunes(app):
             return counts, hrly, raw_left
 
     counts, hrly, raw_left = asyncio.run(_go())
-    assert counts["buckets_rolled"] == 1
-    assert counts["raw_rows_rolled"] == 2
+    # The counts span the whole table, which other tests share (docker
+    # snapshots are recorded at their own, older times, #5752): this host's
+    # bucket is checked exactly below.
+    assert counts["buckets_rolled"] >= 1
+    assert counts["raw_rows_rolled"] >= 2
     assert len(hrly) == 1
     assert abs(float(hrly[0][0]) - 15.0) < 0.001   # avg(10, 20)
     assert hrly[0][1] == 2