As agreed on 26 September 2025 weekly meeting:

We should improve identification of crashes:

1)is truncated xml enough to assume its a crash

2)can we distinguish between the crash and the test executable being terminated by supervising process due to timeout/too long execution, 

3) can we identify between actual crash (SIGKILL, seg fault and other) from other failures - resulting in non 0 exit code and truncated xml. Do we have such information in test results (coin) logs?

4) in case of actual crash is there enough information eg stack trace or other logged information to make it useful for a developer to fix it (in many Jira tickets created by TestResults Metrics Bot we got comments that developers cannot re-create the crash)

5) clarify n_reruns column in crash_tests PSQL -  right now the number varies between 1-90