All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=844607

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=bb1a237740e6d491ec78a17f18b6d490b2139573461e268b957bcf6a05492a37


Bot(s) for this bug's original alert(s):

win
win-32

For this one, the bisect said it could not select a test. The Bisect button is grayed so can't trigger manually either. Are these tests no longer? +grt@ to confirm, +sullivan@ in case we need to remove alerts of this kind.

A couple of things happening here:
* The sizes tests are run on the builders, not the testers, so they are not bisectable.
* We build every chromium revision on the builders, so generally this isn't an issue.
* If there's just a single revision in the CL range, the dashboard is supposed to auto-assign the bug to the revision author.
* In this case, the dip after the original bump seems to have confused the dashboard into putting the alerts in the wrong place. I think this also confused it into thinking there's not a single revision responsible?

I'll fix it, but the summary is that this should be assigned to thakis because all the alerts came from r558402, "Reland "Reland "win: Link with lld instead of link.exe by default"""

thakis, is the binary size increase expected?

It's not expected, but we learned about it when we landed the switch an earlier time. The switch is currently reverted, and we've since fixed the issue upstream and it won't reappear the next time we reland the switch :-)