Check use of `benchmark_dataset`

We are using this as a key (with "true") to indicate which dataset is the reference.

Is it clear that this means "this one in the benchmark"? Or could this mean "compare this [against the benchmark]"?

Would "is_benchmark" or "is_reference" be any clearer?

Start by locating every use of the `benchmark_dataset` key in the CMEW workflow and determine whether it marks the reference dataset. Review the proposed names with maintainers; done means the meaning is unambiguous and all affected usages consistently use the agreed name.

Written by the indexing model from the issue text.