Update 2-ply-back continuation history

Elo   | 0.72 +- 0.62 (95%)
SPRT  | 10.0+0.10s Threads=1 Hash=16MB
LLR   | 2.11 (-2.94, 2.94) [0.00, 2.00]
Games | N: 332766 W: 80290 L: 79598 D: 172878
Penta | [1333, 39748, 83523, 40452, 1327]

Bench 7141438

Bump version to 1.21

Bench 7190160

Bump version to 1.20

Bench 6215176

Bump version

Bench 8157335

New neural net

Finetuning previous net with 9.28B positions
(added 1B positions from regular chess games)

First stage: LR=2.0e-4, 27B iterations
Second stage: LR=1.0e-5, 24B iterations

Elo   | 8.46 +- 4.29 (95%)
SPRT  | 10.0+0.10s Threads=1 Hash=16MB
LLR   | 2.94 (-2.94, 2.94) [0.00, 2.00]
Games | N: 12114 W: 3064 L: 2769 D: 6281
Penta | [62, 1369, 2895, 1674, 57]

Elo   | 7.62 +- 4.97 (95%)
SPRT  | 40.0+0.40s Threads=1 Hash=64MB
LLR   | 2.95 (-2.94, 2.94) [0.00, 3.00]
Games | N: 8620 W: 2079 L: 1890 D: 4651
Penta | [14, 924, 2239, 1125, 8]

Bench 10214358

Bump version to 1.17

Bench 5754636

Bump version to 1.16

Bench: 5007136

New neural net

Elo   | 2.34 +- 2.06 (95%)
SPRT  | 10.0+0.10s Threads=1 Hash=32MB
LLR   | 2.96 (-2.94, 2.94) [0.00, 2.00]
Games | N: 51766 W: 12495 L: 12147 D: 27124
Penta | [226, 6066, 12951, 6414, 226]

Bench: 5705199

Bump version to 1.15

Bench: 5564036

Release 1.14

Bench: 7010644