Right. The idea would be to never run Terraform outside of the context of a CICD pipeline, in which case the pipeline would be authenticating using a service principal, not azure-cli. The awkwardness is furthered by how Terraform manages state, it would be difficult to set everything up using a non-CICD module and a backend, and then using that same backend but different module on the CICD pipeline.
My goal is to be able to create a resource group, create some service principals, make them owner of the resource group, and then they're able to set everything Databricks needs up through Terraform. This doesn't seem to be possible right now. I'd love to work with the Databricks Terraform provider group and address these pain points.