Extrapolation not correlation issues.
What Wache could be saying is that you cannot possibly test every scenario, nor equally simulate them. The engineers will have representative data from simulation of a certIn situation to predict how the car should respond in a different event. I think this is where the issue is, and with 3 days alone of testing it's not really a big problem.
In the same way a short stint can be run on a tyre and then that sample data can be use to model how a longer stint would look like, with different variables being changed etc.
Keeping in mind Redbull have an old wind tunnel, they may be forced to extrapolate information from limited primary data collected.
This RB21 leaves a lot of questions in terms of progress, but I wouldnt say it's 4th or 1rst. We really have no idea.
It may just turn on Saturday to be the fastest car after a little tweaking.