Change auto_bisect to post results to perf dashboard.

scripts/slave/recipes/bisect_test/example_bisect_results.expected/failed_direction_test.json

 [
   {
     "cmd": [
       "git",
       "update-ref",
       "refs/heads/master",
       "refs/remotes/origin/master"
     ],
     "cwd": "[TMP_BASE]/bogus_tmp_1",
     "name": "git update-ref"
@@ skipping 150 matching lines @@
       "blob",
       "c001c0de:DEPS"
     ],
     "cwd": "[TMP_BASE]/bogus_tmp_1",
     "name": "Resolving reference range.git cat-file c001c0de:DEPS",
     "stdout": "/path/to/tmp/",
     "~followup_annotations": [
       "@@@STEP_NEST_LEVEL@1@@@",
       "@@@STEP_LOG_LINE@Debug Good Revision DEPS@v8: None@@@",
       "@@@STEP_LOG_LINE@Debug Good Revision DEPS@skia: None@@@",
-      "@@@STEP_LOG_END@Debug Good Revision DEPS@@@",
-      "@@@SET_BUILD_PROPERTY@extra_result_code@[\"B4T_LO_FINAL_CONF\"]@@@"
+      "@@@STEP_LOG_END@Debug Good Revision DEPS@@@"
+    ]
+  },
+  {
+    "cmd": [
+      "cat",
+      "{\"bad_revision\": \"deadbeef\", \"bisect_bot\": \"linux_perf_tester\", \"bug_id\": \"\", \"buildbot_log_url\": \"\", \"change\": null, \"command\": \"src/tools/perf/run_benchmark -v --browser=release smoothness.tough_scrolling_cases\", \"culprit_data\": null, \"fail_reason\": \"The metric values for the initial \\\"good\\\" and \\\"bad\\\" revisions match the expected direction of improvement. Thus, likely represent an improvement and not a regression.\", \"good_revision\": \"c001c0de\", \"metric\": \"mean_input_event_latency/mean_input_event_latency\", \"revision_data\": [], \"score\": 0, \"status\": \"completed\", \"try_job_id\": null, \"warnings\": [\"This is a demo warning\"]}"
+    ],
+    "cwd": "[SLAVE_BUILD]",
+    "name": "Results",
+    "stdout": "/path/to/tmp/json",
+    "~followup_annotations": [
+      "@@@STEP_LOG_LINE@json.output (invalid)@null@@@",
+      "@@@STEP_LOG_END@json.output (invalid)@@@"
     ]
   },
   {
     "cmd": [
       "python",
       "-u",
-      "\nimport shutil\nimport sys\nshutil.copyfileobj(open(sys.argv[1]), sys.stdout)\n",
-      "---bisect results start here---\n\n===== BISECTION ABORTED =====\nThe bisect was aborted because The metric values for the initial \"good\" and \"bad\" revisions match the expected direction of improvement. Thus, likely represent an improvement and not a regression.\nPlease contact the the team (see below) if you believe this is in error.\n\nBug ID: \n\nTest Command: src/tools/perf/run_benchmark -v --browser=release smoothness.tough_scrolling_cases\nTest Metric: mean_input_event_latency/mean_input_event_latency\nGood revision: c001c0de\nBad revision: deadbeef\n\n\n===== WARNINGS =====\nThe following warnings were raised by the bisect job:\n\n * This is a demo warning\n\n\n===== TESTED REVISIONS =====\nDepot Revision Mean Value Std. Dev. Num Values Good?  \n\n\n\n| O O | Visit http://www.chromium.org/developers/speed-infra/perf-bug-faq\n|  X  | for more information addressing perf regression bugs. For feedback,\n| / \\ | file a bug with label Cr-Tests-AutoBisect.  Thank you!"
+      "RECIPE_MODULE[perf_dashboard]/resources/post_json.py"
     ],
     "cwd": "[SLAVE_BUILD]",
-    "name": "Results",
+    "name": "Post bisect results",
+    "stdin": "{\"data\": {\"bad_revision\": \"deadbeef\", \"bisect_bot\": \"linux_perf_tester\", \"bug_id\": \"\", \"buildbot_log_url\": \"\", \"change\": null, \"command\": \"src/tools/perf/run_benchmark -v --browser=release smoothness.tough_scrolling_cases\", \"culprit_data\": null, \"fail_reason\": \"The metric values for the initial \\\"good\\\" and \\\"bad\\\" revisions match the expected direction of improvement. Thus, likely represent an improvement and not a regression.\", \"good_revision\": \"c001c0de\", \"metric\": \"mean_input_event_latency/mean_input_event_latency\", \"revision_data\": [], \"score\": 0, \"status\": \"completed\", \"try_job_id\": null, \"warnings\": [\"This is a demo warning\"]}, \"url\": \"https://chromeperf.appspot.com/post_bisect_results\"}",
+    "stdout": "/path/to/tmp/json",
     "~followup_annotations": [
-      "@@@STEP_LOG_LINE@python.inline@@@@",
-      "@@@STEP_LOG_LINE@python.inline@import shutil@@@",
-      "@@@STEP_LOG_LINE@python.inline@import sys@@@",
-      "@@@STEP_LOG_LINE@python.inline@shutil.copyfileobj(open(sys.argv[1]), sys.stdout)@@@",
-      "@@@STEP_LOG_END@python.inline@@@"
+      "@@@STEP_LOG_LINE@json.output@{@@@",
+      "@@@STEP_LOG_LINE@json.output@  \"status_code\": 200@@@",
+      "@@@STEP_LOG_LINE@json.output@}@@@",
+      "@@@STEP_LOG_END@json.output@@@"
     ]
   },
   {
     "name": "$result",
     "recipe_result": null,
     "status_code": 0
   }
 ]