Lift — more bins at aspect 10

Same Newton $Q=I/D$ and mass-opt cut $Q>10/11$ as q4. The q4 bottleneck was the compact $\gamma$ at $R=10$ , $n=26$ ( $\varphi=0.9091$ , $P$ error $\approx 0.00469$ ), not the cut.

Raising $n$ on the mid-radius Rayleigh:

$R=10$ , $n=28$ , $\varphi=0.9097$ gives $\gamma=0.905348$ , $1/\gamma=1.104547<1.1046$ .
$R=10$ , $n=30$ , $\varphi=0.9103$ gives $\gamma=0.906238$ , $1/\gamma=1.103463<1.1035$ .
Cut $10/11\approx 0.90909>\gamma$ on both rows.

$1{,}073{,}741{,}823$ faces at the winning row, copositive, $420$ residual skips, $\min m^\top Mm>5\cdot 10^{-4}$ . Stdlib rebuild of $A$ matches the stored matrix to $10^{-15}$ .

Same §7 chain. Printed

N_c<1.1035Z+3.941\,Z^{1/3}\qquad(Z\ge 4).

$R\le 9$ with the mass-opt cut cannot beat $1.1057$ . The withdrawn $1.1168$ stays withdrawn. Finite $Z$ and $N_0(Z)-Z$ stay residue.

Replay: ../run_all.sh.