Adding perf test isolate and gn build target for swarming the benchmarks.

chrome/test/BUILD.gn

Index: chrome/test/BUILD.gn
diff --git a/chrome/test/BUILD.gn b/chrome/test/BUILD.gn
index 36693bef5c8c29231f17f40b71b16e3c6b1855b1..55b006dfecfa5ff5f03fef872b2a9ba1eb7ef7c7 100644
--- a/chrome/test/BUILD.gn
+++ b/chrome/test/BUILD.gn
@@ -1715,6 +1715,20 @@ group("telemetry_perf_unittests") {
   ]
 }
 
+group("telemetry_perf_tests") {

[dtu, 2016/08/30 18:19:00]

I think I prefer telemetry_benchmarks?
Our nomenclature is super confusing between run_tests and telemetry_perf_tests
and telemetry_perf_unittests.

[eyaich1, 2016/08/31 14:01:54]

telemetry_benchmarks may be misleading given that other non-telemetry based perf
tests will be run on these swarmed bots.  I would like to keep this target
consistently named here and in the gn_isolate_map and how we reference them in
the json files.  Changing this only here might get confusing on how it is
referenced and I don't want to change this everywhere given it will eventually
be more than just the benchmarks.  

I think telemetry_perf_tests is consistent in naming, ie unittests are reserved
for tests that test the framework.  I think the run_tests and other scripts
should be updated to match this naming convention.

+  testonly = true
+  deps = [
+    "//tools/perf/:perf",
+  ]
+
+  data = [
+    # For isolate contract.

[dtu, 2016/08/30 18:19:00]

Sorry, I don't understand.

[eyaich1, 2016/08/31 14:01:54]

These are dependencies needed for the script that is actually run in the isolate
to execute.  I updated the comment.

+    "//testing/scripts/common.py",
+    "//testing/xvfb.py",
+    "//testing/scripts/run_telemetry_benchmark_as_googletest.py",
+  ]
+}
+
 js2gtest("unit_tests_js") {
   test_type = "unit"
   sources = rebase_path(