Suppress GCC 16's sfinae-incomplete warning

[DeeJayLSP]

Addresses part of #119269

sfinae-incomplete was introduced in GCC 16, which is meant to catch definitions of types that failed to be complete in an SFINAE context.

At the moment, this happens with the STATIC_ASSERT_INCOMPLETE_TYPE() macro followed by completing that type, which is a context we do want the type to be incomplete, meaning the warning is a false positive.

Note: it has to be done with env.AppendUnique instead of adding to common_warnings as cc1 will spam warnings otherwise.

[akien-mga]

It's not a regression, it's a new warning which flags a true positive. We just don't consider this to be a something that should be fixed, as we do it on purpose.

[DeeJayLSP]

Modified the wording to remove "regression" and "false positive".

[DeeJayLSP]

(This is assuming that the warning is suppressable during SFINAE. If not, disregard)

Note that the warning is emitted on the class definitions that were checked by the macro, not on the macro itself.

For example:

STATIC_ASSERT_INCOMPLETE_TYPE(class, RenderingServer);
//...
#include "servers/rendering/rendering_server.h" // Warning will be emitted in class RenderingServer : public Object {

[Repiteo]

Note: it has to be done with env.AppendUnique instead of adding to common_warnings as cc1 will spam warnings otherwise.

This happens because common_warnings is applied to both C++ and C