nightly-checkuri is prone to intermittent failures due to transient network failures, these don't make sense to pre-triage each time we do a nightly build. Their output is useful, however and therefore we should move those to be triggered on a weekly basis.
Lower number of parallel threads to not trigger DoS detection in the upstream servers?
Good idea. nightly-checkuri already sets BB_NUMBER_THREADS = "1", but that's not the case for the checkpkg selftest.
nightly-checkuri is no longer triggered by nightly and has its own weekly trigger on the yocto.io AB cluster.