return in qs when reaching alpha

[Ergodice]

Simplify score improvement in qsearch

Before if we passed alpha but didn't achieve a beta cutoff we updated alpha and kept alpha. This patch makes it so that a value greater than alpha is returned immediately.

Passed STC:
https://tests.stockfishchess.org/tests/view/67383c6a86d5ee47d953ee9a
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 44992 W: 11764 L: 11551 D: 21677
Ptnml(0-2): 146, 5238, 11505, 5471, 136

Passed LTC:
https://tests.stockfishchess.org/tests/view/6738c2ea86d5ee47d953eef8
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 210474 W: 53710 L: 53683 D: 103081
Ptnml(0-2): 123, 23186, 58607, 23183, 138

Bench: 798787

[pb00068]

Wondering if this change has an impact on mate finding.

[peregrineshahin]

Wondering if this change has an impact on mate finding.

I'm also not a fan of this for the same reason, because if we are in check when the window size is the maximum(-inf, inf) and we enter qsearch, we depend artificially on the fact that all the moves will be searched, returning no uncorrect mated, but with this the first move will be searched and we announce that we get mated while we could escape or delay it..

[peregrineshahin]

For reference here is the CI check proving it has 1 wrong mate and many better mates, I forgot that we put a CI test for it
https://github.com/official-stockfish/Stockfish/actions/runs/11944235795/job/33294816309?pr=5687

[Ergodice]

this tweak is problematic for the gain in simplicity so I am closing this pr