Don't implement Copy for types larger than 128 bits.

[qwandor]

No description provided.

[m4tx]

What's the benefit of this? Usually the compiler is smart enough to figure out the type is big and it's cheaper to pass the reference rather than value. You can see in the code here that [u8; 1024] is actually passed by reference and it doesn't matter whether you manually pass reference, and there's no memcpy in the resulting assembly as long as you enable optimizations. Coincidentally, I even answered this very question (whether it makes a difference to pass Copy types by reference) a while ago on Stack Overflow: https://stackoverflow.com/a/79167130

If there's some additional value in this that I don't see right now, I think we should keep the missing_copy_implementations lint, as we already enable large_types_passed_by_value clippy lint anyway (note that it triggers for types larger than 256 bytes by default).

[qwandor]

Okay, dropping this.