This is an artifact of the Getting Started Guide's workflow being baked in to the current code too simplistically (that is, with a less demanding use case than is generally useful). In short, we took a short cut and now need to do something more generalized. I am aware of work that is underway, that will make it possible for users to control which claims are used to satisfy different aspects of a given proof request. This is a necessary requirement before the ecosystem is mature.
Here is some thinking that shows some of the possibilities for semantics of proof requests, and that implies the need for users to be able to control how claims are used to generate proofs. (This doc has a number of problems because it borrows incorrectly from JSON-LD, etc; don't get hung up on the details but rather consider how it implies some requirements for proof generation.) https://docs.google.com/document/d/117cuMhMF_BLMniyCwnd0cu_Uj5H_45loZiqO1a3Ln7U/edit