Contiguous should be derivable for enums when only the first discriminant is an expression

[ds84182]

This is currently rejected because it isn't an integer literal.

Example:

const MY_CONST: u8 = 123;
#[derive(Copy, Clone, bytemuck::Contiguous)]
#[repr(u8)]
enum Foo {
    A = MY_CONST,
    B,
    C,
}

[Lokathor]

Yes, this sounds right.

As usual for proc-macros, I'm not interested in doing this myself, but if someone else submits the PR then we can get it handled and published.