The Idea behind this ticket is to factor out the targeting code to be more stateless and not rely on an cached CRI variable. The idea being that the new Unified Write Executor has a different refresh workflow and doesn't need a CRITargeter to cache a CRI that we will replace when you call refresh on CRITargeter. CRITarger will then become a class that just hang on to a CRI and errors for use in the BWE code, but for targeting it will pass through to common targeting functions used by both

CRITargeter also now has an internal routing context that it owns and will manage as part of the refresh logic, that can also stick with the stateless code and the targeting logic can just take a reference to a routingCtx