Upgrade to a NNUE trained on Berserk Data.

[TheBlackPlague]

This PR switches to a new network trained on Berserk data (more than 1B data points). Thanks to @jhonnold for all the data he provided and continues to provide.

Architecture remains the same:

       IN          ACCUMULATOR                               HIDDEN                                    OUT
 ______________      _______      ______________________________________________________________      _____
| WHITE: (768) | -> | (256) | -> | ClippedReLU(0, 1) -> (256) \                                 |    |     |
|              |    |       |    |                             \                                |    |     |
|              |    |       |    |                              CONCATENATE(ColorToMove): (512) | -> | (1) |
|              |    |       |    |                             /                                |    |     |
| BLACK: (768) | -> | (256) | -> | ClippedReLU(0, 1) -> (256) /                                 |    |     |
 --------------      -------      --------------------------------------------------------------      -----

ELO Difference

TC: 10s + 0.1s

Using UHO_XXL_+0.90_+1.19.epd:

ELO   | 56.18 +- 18.12 (95%)
SPRT  | 10.0+0.10s Threads=1 Hash=16MB
LLR   | 2.99 (-2.94, 2.94) [0.00, 5.00]
GAMES | N: 1048 W: 461 L: 293 D: 294