🔍 PVS-LMR logic improvement

[eduherminio]

Yeah, it searches deeper. but oh, well

1757: 🤦🏽‍♂️
Fix/Improve PVS-LMR logic
+
Increase Sacrifices test depth from 12 to 20

Score of Lynx 1757 - lmr logic vs Lynx 1756 - main: 10 - 100 - 15  [0.140] 125
...      Lynx 1757 - lmr logic playing White: 7 - 46 - 10  [0.190] 63
...      Lynx 1757 - lmr logic playing Black: 3 - 54 - 5  [0.089] 62
...      White vs Black: 61 - 49 - 15  [0.548] 125
Elo difference: -315.3 +/- 79.3, LOS: 0.0 %, DrawRatio: 12.0 %
SPRT: llr -1.76 (-60.8%), lbound -2.25, ubound 2.89

1769 👎🏽
Re-add lack of best move condition:

Score of Lynx 1769 lmr improv vs Lynx 1756 - main: 55 - 90 - 75  [0.420] 220
...      Lynx 1769 lmr improv playing White: 35 - 30 - 46  [0.523] 111
...      Lynx 1769 lmr improv playing Black: 20 - 60 - 29  [0.317] 109
...      White vs Black: 95 - 50 - 75  [0.602] 220
Elo difference: -55.7 +/- 37.6, LOS: 0.2 %, DrawRatio: 34.1 %
SPRT: llr -0.811 (-28.1%), lbound -2.25, ubound 2.89

1767: Fix bug, setting initial reduction to 0 as the logic of this impl requires

8+0.08

Score of Lynx 1767 lmr improv vs Lynx 1756 - main: 3167 - 3019 - 2878  [0.508] 9064
...      Lynx 1767 lmr improv playing White: 2027 - 1059 - 1446  [0.607] 4532
...      Lynx 1767 lmr improv playing Black: 1140 - 1960 - 1432  [0.410] 4532
...      White vs Black: 3987 - 2199 - 2878  [0.599] 9064
Elo difference: 5.7 +/- 5.9, LOS: 97.0 %, DrawRatio: 31.8 %
SPRT: llr 1.75 (60.4%), lbound -2.25, ubound 2.89

40+0.4

Score of Lynx 1782 - lmr vs Lynx 1781 - main: 557 - 444 - 670  [0.534] 1671
...      Lynx 1782 - lmr playing White: 398 - 131 - 307  [0.660] 836
...      Lynx 1782 - lmr playing Black: 159 - 313 - 363  [0.408] 835
...      White vs Black: 711 - 290 - 670  [0.626] 1671
Elo difference: 23.5 +/- 12.9, LOS: 100.0 %, DrawRatio: 40.1 %
SPRT: llr 2.9 (100.3%), lbound -2.25, ubound 2.89 - H1 was accepted