You also need to test non-regression on LTC, even though I don't think it's necessary.
In my opinion, I believe it's best to leave the way it is, because it's kind of warning that the Bitboard will be changed. But I'm not against this change.

@BM123499 I think using pointers where references will do is ugly and error-prone, but I don't want to revive the reference-vs-pointer discussion here.

I was saving this for a "Small cleanups" PR where usually no regression tests are run at all, but I was getting tired of waiting (and rebasing) so I made this PR. I'll run an LTC test if people think it's required, but I personally think it's a waste of resources.

I have compiled this patch, calculated the md5 hash of the executable binary and compared it to the md5 hash of master: they are the same, so the compiled binaries are the same at least for my compiler (gcc-10 on Mac). So yes, LTC is unnecessary in that case :-)

Patch looks good to me.

@gvreuls I understand your point of view, but I believe that the absence of signally that the value will be change inside another function has more change of a new dev misread what is going to happen.
Stockfish is a international code, we have a lot of code style here, so I believe that both ways to write is valid. We learn on different ways how to code and none is wrong.
So if most people are aligned with your way of reading the code, then let's change. I'm just remarking my point of view.

@BM123499 I think most C++ programmers expect that a Bitboard& function argument will be modified by the function, if it won't be modified they'd expect a const Bitboard& signature (or in case of bitboards a simple pass by value Bitboard) instead.

I appreciate and understand your point of view, don't worry. I think it's related to what code you read/write most often, C or C++.

Passing a pointer also doesn't tell the whole picture on the callsite, for all we know it could be a const Bitboard*. I think this is a good change. The name of the function should reflect whether it modifies any arguments; pop_* is enough to signify that imo.

Merged via ec42154, thanks! :-)