Manipulate Arrays as if they were vectors!

This crate exposes a trait that allows manipulating arrays in a vec-like fashion.

Alternatives like ArrayVec operate over a [MaybeUninit<T>; N]-like data structure and panic if the size is overflown.

Instead, the point of this crate is allowing to "resize" arrays. Internally it just memcpy()s stuff around, so a linked list might be preferable as writing a pointer is much faster than copying memory. However, maybe LLVM is able to ultra-optimize this, idk.

As I said, if you heavily depend on this crate, probably a linked list will be much more performant, but this crate is still very useful for one-time operations, operations where using a linked list will give more problems than solutions, coercing arrays or devices where you can't allocate.

This crate works & performs exceptionally well when you have an array that you need to manipulate but you still need to use a fixed array later and not a Vec. In that situation you just avoided 1 malloc() & 1 memcpy() best-case scenario and n malloc()s, 1 memcpy() and a conditional worst-case scenario.

Merging into core

This crate depends on the experimental (not complete) feature generic-const-exprs, so the Pre-RFC is postponed until it doesn't get to a more mature state.