I understand in general the constraints of the RIM model and the steep learning curve of v3, which has led to many of its failures. Although FHIR has clear strengths in scalability and application to mobile devices, is it being developed as a total replacement or alternative?
If its an alternative, what are strictly the different uses for each?
There's two parts to the answer, for v3 and CDA
v3 is (was) a general methodology for defining a set of well-described exchangeable content, and for creating exchange packages derived from them. v3 was primarily used by National Programs (UK, Canada, Netherlands) but a little use happened elsewhere
FHIR leverages the well described exchangeable content, but replaces the exchange packages part with a RESTful API suitable alternative. FHIR has been developed as a total replacement for v3, though there's still a lot to do before the task is done. There's still a little bit of new development of exchange content being done in v3, but it is tailing off.
There was a specific subset of the v3 exchange packages that was much more widely adopted - documents (CDA and SPL). The FHIR architecture offers the ability for FHIR to completely replace both CDA and SPL, but both these have considerable market investment and ongoing development and adoption, and it seems quite unlikely that FHIR will quickly replace them. Instead, it seems more likely to me that FHIR will gradually be adopted in parallel to CDA for data centric use cases, while CDA will remain the focus for document exchange (at least for the next few years).
Eventually, CDA and/or SPL may publish updates that replaces their existing technical base with a FHIR based alternative that meets the same use case, but no one is in any hurry to do that now (if at all, for SPL). It will be a question of when the apparent technical advantages of a new approach that is consistent with an increasingly widely adopted data exchange format justifies the price of changing.