Labeling bisect/perf try jobs with a prominent null step.

scripts/slave/recipes/bisection/desktop_bisect.expected/perf_tryjob_failed_test.json

 [
   {
     "cmd": [
       "python",
       "-u",
       "RECIPE_MODULE[depot_tools::bot_update]/resources/bot_update.py",
       "--master",
       "tryserver.chromium.perf",
       "--builder",
       "linux_perf_bisect",
@@ skipping 177 matching lines @@
     ],
     "cwd": "[SLAVE_BUILD]/src",
     "name": "git diff to analyze patch",
     "stdout": "/path/to/tmp/",
     "~followup_annotations": [
       "@@@STEP_LOG_LINE@files@tools/run-perf-test.cfg@@@",
       "@@@STEP_LOG_END@files@@@"
     ]
   },
   {
+    "cmd": [],
+    "name": "***PERF TRYJOB***"
+  },
+  {
     "cmd": [
       "python",
       "-u",
       "RECIPE_MODULE[build::perf_try]/resources/load_config_to_json.py",
       "--source",
       "[SLAVE_BUILD]/src/tools/run-perf-test.cfg",
       "--output_json",
       "/path/to/tmp/json"
     ],
     "name": "load config",
@@ skipping 450 matching lines @@
       "@@@STEP_FAILURE@@@"
     ]
   },
   {
     "name": "$result",
     "reason": "Failed to upload result to Perf Dashboard. Error response: None",
     "recipe_result": null,
     "status_code": 1
   }
 ]