The definition contains the following wording:

For a non-subject-based measure, this would be the type of elements in the population (such as encounters or procedures).

The objection is to to the wording "non-subject-based measure". All FHIR measures have a subject, but the subject is not required to be given if the subject is Patient. Per the subject definition:

The intended subjects for the measure. If this element is not provided, a Patient subject is assumed, but the subject of the measure can be anything.

Recommend updating "non-subject-based measure" to "measure that uses the default subject of Patient".