The timeseries::bucket_catalog::conflictsWithReopening function does not return true for frozen buckets even though logically they do conflict with reopening. The reason that it doesn't return true in this case is because its callers often use it to determine whether to return a write conflict, but a write conflict is not the correct action for a frozen bucket because its state will never change. We may want to adjust the naming/function usage here in order to improve the semantics.