As we're moving to scheduling releases in each cluster, it will become
useful to split choosing which clusters to schedule a release on from
the actual scheduling, as choosing the clusters can only happen once,
while we'll schedule the release once per cluster.
As we're moving to scheduling releases in each cluster, it will become useful to split choosing which clusters to schedule a release on from the actual scheduling, as choosing the clusters can only happen once, while we'll schedule the release once per cluster.