Sometimes change the (materialist, positional) balance

[snicolet]

Our new nets output two values for the side to move in the last layer.
We can interpret the first value as a material evaluation of the
position, and the second one as the dynamic, positional value of the
location of pieces.

This patch changes the balance for the (materialist, positional) parts
of the score from (128, 128) to (121, 135) when the piece material is
equal between the two players, but keeps the standard (128, 128) balance
when one player is at least an exchange up.

Passed STC:
LLR: 2.93 (-2.94,2.94) <-0.50,2.50>
Total: 15936 W: 1421 L: 1266 D: 13249
Ptnml(0-2): 37, 1037, 5694, 1134, 66
https://tests.stockfishchess.org/tests/view/60a82df9ce8ea25a3ef0408f

Passed LTC:
LLR: 2.94 (-2.94,2.94) <0.50,3.50>
Total: 13904 W: 516 L: 410 D: 12978
Ptnml(0-2): 4, 374, 6088, 484, 2
https://tests.stockfishchess.org/tests/view/60a8bbf9ce8ea25a3ef04101

closes #3492

Bench: 3856635

[vondele]

as we use the output of the network in the pure form to check the implementation (see tools branch stockfish/script/cross_check_eval.py), it would probably be a good idea if we give the evaluate function a bool argument that is false by default, and only enables this from the adjusted_NNUE function.

[vondele]

I've given that a try on the above mentioned branch, that works for me.

[snicolet]

I'm fine with it :-)

[snicolet]

Another idea which could be useful in the future is to introduce a maximum bound for the positional twist, which could be adjusted via a private UCI parameter.

That way we could use the same binary twice during game generation, one player with a big twist and the other one with a zeroish twist, to have games with clashing styles using the same binary.