Project X would be ticking along nicely if it wasn't for sodding local government agencies
Anyhoo.. having a break from that, I'll throw my 2p into the ring.
If there is a problem (and I'm not convinced that there is) then it is that some people are shooting in a lower grade than perhaps they ought to be.
We can come up with a complex solution and check that it removes all of the outliers with the current data set, but that's certainly no guarantee that there won't be outliers in future.
Whatever normalisation function we use there will always be outliers.
My suggestion would be to introduce some form of manual override to the current system.
So perhaps a committee at regional/national level that can make a decision to change an individual's grade if they believe that the calculated grade is incorrect for whatever reason.
That of course is making the assumption that there actually is a problem that's worth solving.
The grading system is NOT a measure of current performance, it is a prediction of future performance and as such it will inevitably make an incorrect prediction from time to time.