Make error handling consistent when iterating over clusters

When loading a sample bucket fails against one cluster in the list of identifiers this does not cause the entire command to fail, instead the results of each attempted load are reported. This is not the case will all the other bucket commands, these return an error when one of the clusters returns an error. We should make the behaviour consistent and have all commands that take a list of cluster identifiers act the same as `buckets load-sample`.

Start by locating the `buckets load-sample` implementation and the other bucket commands that accept lists of cluster identifiers. Compare how each command handles a failed load, then make their behavior consistent with the chosen convention and verify that every listed cluster is handled the same way.

Written by the indexing model from the issue text.