Assert moveCountPruning properly

[MinetaS]

moveCountPruning should not be set to false again once it is set to true, otherwise it will break the internal logics of MovePicker. There have been a few tests which overlooked this.

Examples:
https://tests.stockfishchess.org/tests/view/671e7c0486d5ee47d953d226 https://tests.stockfishchess.org/tests/view/66a1de7b4ff211be9d4eccea

No functional change

[mstembera]

Would

moveCountPruning = moveCountPruning || moveCount >= futility_move_count(improving, depth);

not do the same thing?

[MinetaS]

Would

moveCountPruning = moveCountPruning || moveCount >= futility_move_count(improving, depth);

not do the same thing?

That's the "most correct" form that we should have, however this simplification was merged before: #5213

There are two options ...

1. Introduce a strict assertion as this PR presents, or
2. Enact a guideline that forbids removing the explicit check to prevent such PRs from being merged.

[Disservin]

what about "fixing" the movepicker instead ? i.e. add a function skip_quiets(), which sets an internal bool to true and remove the parameter from next_move, would prevent any code assertion and additional maintenance from a pr stand point?

[MinetaS]

what about "fixing" the movepicker instead ? i.e. add a function skip_quiets(), which sets an internal bool to true and remove the parameter from next_move, would prevent any code assertion and additional maintenance from a pr stand point?

That sounds like a really good idea.

[MinetaS]

Closing as per the follow-up PR: #5661