We have checked with our team and they have conveyed it is possible to use but they are working on the documentation. We are trying to find more info regarding this.
Here is one of the scenario they shared - One is moving on-prem data over IRT copy pipeline then invoking bulk loaders (for ndjson and/or FHIR bundles) or you can place the bundles in a Gen2 storage account hierarchy and they will be loaded into FHIR Server this is a function not an ADF workflow.
Regards
Navtej S