Make sure stats updates are never bigger than maximum value of stats.

[Vizvezdenec]

In current master you always need to keep track if you may update stats with values too big for them.
This patch makes sure that even if you pass such update it would be lowered to value low enough, so is a QoL change.
Passed non-regression bounds :
https://tests.stockfishchess.org/tests/view/65ef2af40ec64f0526c44cbc
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 179232 W: 46513 L: 46450 D: 86269
Ptnml(0-2): 716, 20323, 47452, 20432, 693
non-functional change

[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 8299644774 / attempt 1)

[Disservin]

you also add to include #include <algorithm> again :d

[Viren6]

I think you should keep the assert. While the clamp guards the overflow, its still likely not intended behaviour.

[Vizvezdenec]

Removing assert and thus allowing to sometimes pass overflowing bonuses is the sole purpose of this patch.
It's pretty important for patches like this https://tests.stockfishchess.org/tests/view/65eeeee00ec64f0526c44832 or patches like this https://tests.stockfishchess.org/tests/view/65ef5ed70ec64f0526c450af where with master you need to calculate "can this cause an overflow" while patch makes everything bulletproof.
Also master way is extremely inconvenient since you pass bonuses in search but max value is listed in movepicker so you need to constantly check different files to see if you did smth wrong.

[Disservin]

Yeah I agree with viz, I don't see much value of keeping the assert while adding the clamp, will just make a fishtest patch pass but then fail our ci.