I am more familiar with the basic research phase but in my experience very few groups actually treat model choices as an engineering tradeoff. Labs usually write grants centered around whatever model organism(s) the lab is already used to, and it's rare to see a proposal that doesn't use one of a few "accepted" models. I've never seen anyone sit down and have a discussion at the proposal stage about this, to actually seriously consider alternative setups or even just acknowledge why they are choosing this same mouse again. The tradeoff defense comes after the fact, and really only if someone else brings it up.

I think it is in part a symptom of the cost of research causing labs to be centered moreso around a methodology than a question. Having some labs like this isn't a problem, but when an entire field starts to trend this way, and almost everything that gets funding chooses a particular side of a tradeoff (that isn't super well quantified to begin with), then I think we could end up missing a lot.