Clusterfuzz fails to resolve Chrome bug due to spurious "known crash revision"

[dmcardle]

On https://crbug.com/1252658, I merged a fix at revision r925365. I'm pretty certain that this fixes the crash because I added a unit test based on the Clusterfuzz minimized input.

Clusterfuzz failed to mark the issue fixed:

[2021-09-28 18:09:29 UTC] clusterfuzz-windows-nbh6: Progression task errored out: Known crash revision 925641 did not crash.

Strangely, this known crash revision comes after my fix (r925641 > r925365). I'm not clear how this happened.

timothygu mentions in comment 12 that Clusterfuzz has misattributed crashes to this testcase before.