In null move pruning, we only really care about a lower bound. If this exceeds beta, then great, we have a cutoff. The exact value in [alpha, beta] is not that important.
Coverage remained the same at 86.027% when pulling 59403541ee77a99da38f33f6e7ef112aaeba3ffe on null-move-pruning into 03c4d3479284b7719571f403434a32d67a4d5e81 on master.
Merging without play strength check because this is the correct way. Even if by chance this reduces strength some, we would rather have a correct engine than fast and incorrect (unstable).
In null move pruning, we only really care about a lower bound. If this exceeds beta, then great, we have a cutoff. The exact value in [alpha, beta] is not that important.