Many thanks for the report! We will investigate this.

Chan, assign to you as this is test failure, and the original analysis result is not shown due to the failure grouping.

I checked on the analyses and it appeared that 2 independent try jobs identified the same CL as culprit. I think there are 2 places we need to improve:
1. We need to fix the grouping. In this case, at least one analysis in group has identified the test to be flaky, 
  a. in this case we may consider increase the iterations in swarming reruns for the other analyses in group to confirm if the failures are flaky
  b. evaluate the possibility of the same test is flaky on some builds and reliably fails on some other builds at the same time. If the possibility is very low, I may consider treating all failures in a group as flaky if any of the results is flaky.

2. For both analyses, in the try jobs the test on the revision before the culprit is skipped. We should consider always run the previous revision to confirm the culprit is correct.

Make sure you follow the status of 716594. Latest theory is a potential difference in bot config, which would throw off find-it.