i used 1e8 first and got wa14
latter used n*(sum of edge lengths) and got ac.

i found that for test 14 the curve goes up and at r=200.75 touches x axis then goes up until
x=around 450 then down again with asymptote as x axis. so i tried to use tenary search to find upper bound but it does not always work. why n*sum of lengths work? is it weak tests or can somebody prove optimal upper bound?

please, help me understand.
thanks in advance.