Proposal: move field declaration align syntax to field name

[Snektron]

In the current Zig grammar, fields declarations can be forced to arbitrary alignment by using the align attribute:

const value align(4) = false;

When an explicit type is given, the align is specified after the type:

const value: bool align(4) = false;

With this grammar, it looks like align is actually part of the type expression, which led me to falsely believe that

const A = struct { x: u32 } align(4);

is valid syntax.
I propose to instead place the align attribute after the type name to improve readability:

const value align(4): bool = false;

Another option would be to actually move the align specifier to the type expression, and have TypeName align(expr) be valid syntax. I think in that case it would be clearer to have the alignment before the type name though:

const value: align(4) bool = false;
const A = align(4) struct { x: u32 };

[Snektron]

I realized there is a similar problem with linksection, as it is currently also specificed after the type.

As for readability, someone commented about this being less readable because the attributes are now specified before the type. A solution to that would be to leave the attributes behind the type, but add some syntax to be able to more clearly separate where the type ends and where the attribute starts, which is also proposed in #5239 and #4285.