The image used in the failing run (registry.svc.ci.openshift.org/ci-op-hffv0hgs/stable@sha256:09242c40896e3d92888ee8b8c4387f942cb5001e0ce56e593b80343209871263) was definitely using non-master code, but the release-4.4 branch of cluster-ingress-operator seems to be up to date with master for over a week, so the next line of investigation might be looking into what's going on with the 4.4 image build pipeline (or whether the test run was actually referencing the correct image in the first place).
The image has since been fixed.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2020:0062