Add test specs for TSan bots on the MFYI bots

testing/buildbot/chromium.memory.fyi.json

 {
   "Linux MSan Tests": {
     "gtest_tests": [
       "accessibility_unittests",
       "app_list_unittests",
       "app_shell_browsertests",
       "aura_unittests",
       "base_unittests",
       "cacheinvalidation_unittests",
       "cast_unittests",
@@ skipping 113 matching lines @@
   },
   "Linux ChromeOS MSan Browser (2)": {
     "gtest_tests": [
       {"test": "browser_tests", "shard_index": 1, "total_shards": 3}
     ]
   },
   "Linux ChromeOS MSan Browser (3)": {
     "gtest_tests": [
       {"test": "browser_tests", "shard_index": 2, "total_shards": 3}
     ]
+  },
+  "Linux TSan Tests": {
+    "gtest_tests": [
+      "accessibility_unittests",
+      "app_list_unittests",
+      "aura_unittests",
+      "base_unittests",
+      "cacheinvalidation_unittests",
+      "cast_unittests",
+      "chromedriver_unittests",
+      "components_unittests",
+      "content_unittests",
+      "crypto_unittests",
+      "dbus_unittests",
+      "device_unittests",
+      "display_unittests",
+      "events_unittests",
+      "extensions_unittests",
+      "gcm_unit_tests",
+      "gfx_unittests",
+      "google_apis_unittests",
+      "gpu_unittests",
+      "ipc_mojo_unittests",
+      "ipc_tests",
+      "jingle_unittests",
+      "media_unittests",
+      "mojo_common_unittests",
+      "mojo_public_bindings_unittests",
+      "mojo_public_environment_unittests",
+      "mojo_public_system_unittests",
+      "mojo_public_utility_unittests",
+      "mojo_system_unittests",

[jam, 2015/02/05 21:12:06]

take out the mojo tests (other than mojo_common_unittests which is testing code
in chromium). If we want these tests to be run under tsan, we should add a
builder on that waterfall.

[Alexander Potapenko, 2015/02/06 12:35:07]

I've copied these from "Linux MSan Tests" above.
Will remove them for now.

+      "net_unittests",
+      "ppapi_unittests",
+      "printing_unittests",
+      "remoting_unittests",
+      "sandbox_linux_unittests",
+      "skia_unittests",
+      "sql_unittests",
+      "sync_unit_tests",
+      "ui_base_unittests",
+      "ui_touch_selection_unittests",
+      "unit_tests",
+      "url_unittests",
+      "views_unittests",
+      "wm_unittests"
+    ],
+    "disabled_tests": {
+      "app_shell_browsertests": "http://crbug.com/455633",
+      "cc_unittests": "http://crbug.com/437454",
+      "content_browsertests":
+        "Too many errors, clean browser_tests first. http://crbug.com/368525",
+      "interactive_ui_tests": "http://crbug.com/455679"
+    }
+  },
+  "Linux TSan Browser": {

[jam, 2015/02/05 21:12:06]

The model now is to have one bot which runs the tests and which uses swarming.
less column on the waterfall is better :)

any reason why these aren't using swarming?

[Alexander Potapenko, 2015/02/06 12:35:07]

I was going to first mimic the current non-recipes configuration and then
improve it gradually.
But you're right, let's just throw in some swarming.

+    "gtest_tests": [
+      {"test": "browser_tests"}
+    ]
   }
 }