[perf] Prepare benchmarks for memory metric normalization

tools/perf/benchmarks/v8.py

Index: tools/perf/benchmarks/v8.py
diff --git a/tools/perf/benchmarks/v8.py b/tools/perf/benchmarks/v8.py
index a0f9aad8aad7b7f5c47a04e4235a6b57e88ece67..775a6f7dcfc737e3f8126ad9c518bd893e7dc691 100644
--- a/tools/perf/benchmarks/v8.py
+++ b/tools/perf/benchmarks/v8.py
@@ -227,7 +227,11 @@ class _V8MemoryAndCodeSizeBenchmark(perf_benchmark.PerfBenchmark):
     # is able to cope with the data load generated by TBMv2 metrics.
     if 'memory' not in value.name:
       return True  # Keep all non-memory values.
-    if 'subsystem' in value.name and 'v8' not in value.name:
+    # TODO(petrcermak): Remove the 'subsystem' disjunct once
+    # https://codereview.chromium.org/2018503002/ lands in Catapult and rolls
+    # into Chromium.
+    if (('subsystem' in value.name or 'reported_by_chrome' in value.name) and
+        'v8' not in value.name):
       return False  # Drop non-V8 values reported by Chrome.
     # Keep dump counts and average+max of process counts, vmstats and v8.
     return not cls._IGNORED_V8_STATS_RE.search(value.name)