Different eval parameters for smallnet and psqtOnly

[gahtan-syarif]

Gives different eval scaling parameters for the three different nets (bignet, smallnet, psqtOnly smallnet).

Edit: Special thanks to maintainer @Disservin for making the code more easily readable by refactoring the code to use lambda functions.

Passed STC:
https://tests.stockfishchess.org/tests/view/65f4b0020ec64f0526c4a3bd
LLR: 2.96 (-2.94,2.94) <0.00,2.00>
Total: 168064 W: 43507 L: 42987 D: 81570
Ptnml(0-2): 662, 19871, 42445, 20393, 661

Passed LTC:
https://tests.stockfishchess.org/tests/view/65f6be1a0ec64f0526c4c361
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 162564 W: 41188 L: 40604 D: 80772
Ptnml(0-2): 120, 18112, 44216, 18732, 102

Bench: 2113576

[github-actions]

clang-format 17 needs to be run on this PR.
If you do not have clang-format installed, the maintainer will run it when merging.
For the exact version please see https://packages.ubuntu.com/mantic/clang-format-17.

(execution 8339489068 / attempt 1)

[vondele]

I suggest you make a function for

        // Blend optimism and eval with nnue complexity and material imbalance
        optimism += optimism * (nnueComplexity + std::abs(simpleEval - nnue)) / 513;
        nnue -= nnue * (nnueComplexity + std::abs(simpleEval - nnue)) / 32395;

        npm = pos.non_pawn_material() / 64;
        v   = (nnue * (919 + npm + 11 * pos.count<PAWN>()) + optimism * (145 + npm)) / 1036;

        // Damp down the evaluation linearly when shuffling
        shuffling = pos.rule50_count();
        v             = v * (178 - shuffling) / 204;

and call it with different parameters, instead of copying it 3 times.

[Disservin]

Yeah, probably make small lambda function instead of a free function to avoid passing the nnue and other variables in the function call.

[gahtan-syarif]

refactored using arrays, special thanks to @cj5716 for the suggestion.

Edit: seems to fail the test for riscv64