Improve usage of repr(packed)

[ghost]

Close #38.
We occasionally userepr(packed) when the struct is unaligned. When we use it, I added a comment explaining why. In addition, we derive Copy and Clone to allow the compiler to copy the fields to locals.

[robert-w-gries]

Looks good to me. Will hold off on merging for a day in case @IsaacWoods wants to review, as I've been less involved in the project lately.

[VictorKoenders]

Maybe outside of the scope of this PR, but should there be compile-time validations or tests to make sure the structs are the right size?

In my project i have a couple of

fn _test_size() {
    unsafe {
        core::mem::transmute::<[u8; 56], Hpet>([0u8; 56]);
    }
}

Which give a compile-time check to validate the Hpet struct is still 56 bytes. Of course this can also be done in a #[test].

Enforcing this would make sure that changes by humans or compilers will get noticed instead of loading wrongly-sized structs from memory pointers.

[ghost]

@VictorKoenders I completely agree. I will add those tests to this PR.

[robert-w-gries]

Thanks for adding the tests