assert(false) should be exempt from the assert message check

[Moth-Tolias]

since assert(false) has the specific meaning of "unreachable code", it should be exempt from the assert message check as to add one would be needless repetition.

[VPanteleev-S7]

I think the same rationale for why it makes sense to warn on missing messages for assert(condition) also applies to assert(false). In both cases, in debug builds, you'd get a nondescript "Assertion failure" message, which the linter rule exists to avoid. The only practical difference is what happens in -release builds, but that seems orthogonal to the aspects concerning the linter.