[ews] re-run-layout-tests should use results database instead of doing clean-tree runs. For example in https://ews-build.webkit.org/#/builders/10/builds/167694 layout-tests step had 4 failures and 3 of them were pre-existing as per results db (see https://ews-build.webkit.org/#/builders/10/builds/167694/steps/16/logs/stdio), but even though the fourth one (streams/readable-stream-default-controller-error.html) wasn't failing in results db, it was actually flaky, and it did not fail in re-run-layout-tests step (https://ews-build.webkit.org/#/builders/10/builds/167694/steps/22/logs/test-failures doesn't have that test failure). EWS did run-layout-tests-without-change (after compiling webkit) and reached this conclusion, but all that extra compilation and run-layout-tests-without-change could have been avoided, if re-run-layout-tests consulted results database.
<rdar://problem/102764839>
Pull request: https://github.com/WebKit/WebKit/pull/6917
This is also needed in cases like https://ews-build.webkit.org/#/builders/10/builds/167706 where the first layout-test run failed to run because of some reason (e.g.: infra issue).
Committed 257117@main (a6df5e4d036f): <https://commits.webkit.org/257117@main> Reviewed commits have been landed. Closing PR #6917 and removing active labels.
More examples where this change would have helped: https://ews-build.webkit.org/#/builders/60/builds/44663 (flaky failure in first run) https://ews-build.webkit.org/#/builders/10/builds/167706 (first run failed due to infra issue)
Restarted EWS to pick up this change.