All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=874882

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=98925450aa541b178ece5005b86209bc88f258f1176e70d1ecb01db41a731bee


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

mac-10_12_laptop_low_end-perf

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/1484ac56640000

📍 Couldn't reproduce a difference.
https://pinpoint-dot-chromeperf.appspot.com/job/1484ac56640000

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/16a1d234640000

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/16a1d234640000

Roll back code that serialized at the CSS block level due to performance. by dmazzoni@chromium.org
https://chromium.googlesource.com/chromium/src/+/b53fffcc225fd0f37ab7f56ba9f075f319d5e439
2.061e+07 → 3.061e+07 (+9.994e+06)

Understanding performance regressions:
  http://g.co/ChromePerformanceRegressions

 Issue 875333  has been merged into this issue.

 Issue 875332  has been merged into this issue.

Three alerts so far, but all on Mac. I wonder if that's relevant.

It's also weird because this is a revert of a change that regressed a performance test: see http://crbug.com/865621

Closing as uninteresting. Only shows up on some bots, and none of the other fluctuations in memory usage recently seem accessibility-related.