From fast routines to implementation of depth permutation parameter...
So your remark (John) is that it is slow...
Well, it can always be slower, lol...
But in fact, I am satisfied now with this depth version... It should yield the same results as previous with depth=3, but if you put a little higher (for ex. 6), IMO it will be very close to exact solution, no matter how long does it take to finish...
I've tested on my tricky example with 11 points and it did correct at about 40 seconds with my PC - this is still faster than doing full permutations... In routine I hard coded - preferable number 3, but maybe it's actually 6, if you have more time...
Regards, M.R.