Remove noexcept(false) annotations

[etremel]

According to everything I can find on the Internet (such as this and this), the noexcept(false) annotation on a function does nothing. Therefore, we should remove it from the Derecho codebase to reduce unnecessary "noise" in the source code and potential sources of confusion for other developers who use our library (they might wonder why some functions are declared noexcept(false) and some are not, until they figure out that not writing noexcept(false) is exactly the same as writing it).

[songweijia]

I agree to remove noexcept(false) since it does not take effect. I do think C++ should come with a non-parameterized noexcept syntax.

[etremel]

Oh, C++ does have a non-parameterized noexcept - if you just write noexcept on a function declaration (with no parentheses) it's exactly equivalent to writing noexcept(true). If you want, I can also change all of our noexcept(true) to noexcept.

[songweijia]

I'm not sure if you noticed a compilation warning on in commit 4a402b6. Some functions in C++ are by default evaluated to noexcept(true), for example, the destructors. And in such a case, noexcept(false) makes sense.
"""
non-throwing functions are all others (those with noexcept specifier whose expression evaluates to true as well as destructors, defaulted special member functions, and deallocation functions)
"""
Source link. So I reversed one of such a case in 3d6038f.

[etremel]

Oh, you're right. I didn't notice the compile warning because our build process produces so much output, but yes, a destructor is one case where declaring noexcept(false) actually has an effect.