See discussion of this issue in https://groups.google.com/a/chromium.org/forum/#!topic/chromium-dev/NMPDjCttrFo.

The TL;DR is that tests sometimes fail only in component build due to violations of the ODR rule. When this happens, the test failure isn't detected until the change hits the main waterfall.

This bug exists to track the frequency of this occurrence. As jam@ wrote on the chromium-dev@ discussion:

"I defer to infra team if we have resources to run debug tests on the CQ. I would prefer we avoid release component build since that would be another config.

It would be good to have a bug to track this, to see how often it happens, to balance that with other untested configs. If these failures are rare in general, then it might not be worth a bot on every try run (even though it is frustrating when it happens)"