Bisect performed in https://bugs.chromium.org/p/chromium/issues/detail?id=758852#c15 incorrectly blames CL that improved the metric after the actual regression.

Regression from 65 -> 373 happened in chromium@496563.
Improvement from 373 -> 179 happened in chromium@496661,v8@af91eb85de

Bisect blamed chromium@496661,v8@af91eb85de for 65 -> 179 "regression".

Perf regression found with culprit

Suspected Commit
  Author : Michael Hablich
  Commit : af91eb85de58d9205f029da39662740ca7c222f2
  Date   : Wed Aug 23 10:18:57 2017
  Subject: Version 6.2.337.1 (cherry-pick)

Bisect Details
  Configuration: mac_pro_perf_bisect
  Benchmark    : v8.runtimestats.browsing_desktop
  Metric       : GC:duration_avg/browse_media/browse_media_flickr_infinite_scroll
  Change       : 165.04% | 65.4935 -> 173.5815

Revision                           Result                  N
chromium@496123                    65.4935 +- 47.419       6      good
chromium@496416                    74.128 +- 15.4836       5      good
chromium@496563                    373.369 +- 42.1095      5      good
chromium@496636                    379.706 +- 143.53       7      good
chromium@496655                    346.358 +- 139.019      6      good
chromium@496660                    352.296 +- 89.9254      6      good
chromium@496661                    356.544 +- 103.89       5      good
chromium@496661,v8@af91eb85de      179.449 +- 52.229       6      bad       <--
chromium@496662                    153.876 +- 30.0466      6      bad
chromium@496664                    175.397 +- 49.3009      6      bad
chromium@496673                    174.572 +- 49.1056      5      bad
chromium@496709                    173.582 +- 45.7311      6      bad