Only time I ever did this, was from the IC design side and
we did it by hand (with a bunch of back-and forth re
arrangement, between the board guys at the customer
and myself).
So that last bit, you might not gain the advantage you think
from a simple automation.
And even with a few hundred balls you might well get it done
faster by hand if you started now, than if you wait for the
magical contrib widget to surface.