And unlike kit designed for the purpose
“Unlike certain other more document-centric standards” is a bit of smoke and mirrors. Clever. What has a document model got to do with it? We have IHE XDS standards to support that.
How about the major issue the M$ adding a persistence model to a technology that was not designed for such is shifting the goal posts significantly. Now there is a proprietary model (in SQL) underpinning the previously open messaging architecture. FHIR does not need a persistence layer to support its core aim of technical interoperability / integration. Leave the persistence to more adaptable tech like openEHR which can make allowances for the incomplete nature of the FHIR model. Try maintaining semantic coherence with endless extensions, no governance on profiles and no reference model. No really, do try.
Good luck...