CQF Questionnaire has been removed in FHIR R5 and has been acknowledge by CDS to be a non-preferred/incorrect way of defining Questionnaires.  (Going forward, CDS is leveraging SDC for handling form population and other logic.)  My understanding was that DTR was moving to comply with SDC (which means using the initialValueExpression, calculatedValueExpression and other extensions).  However, the wording is unclear about expectations.  As well, the spec has two Questionnaire profiles - one based on CQF and one based on SDC.  The SDC one seems to be far more complete, so it seems like the CQF one should be removed, as should the reference to CQF itself.